AIR-CONDITIONING CONTROLLER AND AIR-CONDITIONING SYSTEM

Description

TECHNICAL FIELD

The present disclosure relates to an air-conditioning controller that controls an air-conditioning apparatus and also to an air-conditioning system.

BACKGROUND ART

Some technology for controlling the heating energy environment within an air-conditioned space based on thermo-fluid analysis has been proposed (see Patent Literature 1, for example). This technology is to conduct simulation based on thermo-fluid analysis and thereby to optimize air conditioning control based on the simulation result.

CITATION LIST

Patent Literature

• Patent Literature 1: Japanese Unexamined Patent Application Publication No. 2014-31899

SUMMARY OF INVENTION

Technical Problem

However, simulation based on thermo-fluid dynamics usually takes time to obtain an optimal solution. Additionally, the technology disclosed in Patent Literature 1 is based on the transition from a solution that is obtained by conducting optimization in advance to the next optimal solution. This may make a user feel uncomfortable for a period from a certain state, such as the time of startup, until a control operation based on the optimal solution is executed.

The present disclosure has been made to solve the above-described problem. It is an object of the present disclosure to provide an air-conditioning controller and an air-conditioning system that can speedily improve the comfort level of a user.

Solution to Problem

An air-conditioning controller according to an embodiment of the present disclosure is an air-conditioning controller configured to control an air-conditioning apparatus that performs air conditioning for an air-conditioned space. The air-conditioning controller includes an obtainer, an indoor model constructor, a coupled analysis unit, a control target determiner, and a commander. The obtainer is configured to obtain room shape information indicating a shape of the air-conditioned space and air state information indicating a state of air in the air-conditioned space. The indoor model constructor is configured to construct an indoor model used for CFD simulation, based on the room shape information and the air state information. The coupled analysis unit is configured to execute the CFD simulation based on the indoor model to derive a state quantity indicating the state of air at each time point within a predetermined first time range. The coupled analysis unit is configured to derive a deviation of the state quantity at each time point from a preset target state quantity and integrate the deviations at the individual time points within the first time range. The coupled analysis unit is configured to derive an optimal solution of a control parameter of the air-conditioning apparatus such that an objective function including an integrated value obtained by integrating the deviations is minimized. The control target determiner is configured to determine a control target value, which is a target value of the control parameter, based on the optimal solution. The commander is configured to control the air-conditioning apparatus based on the control target value.

An air-conditioning system according to an embodiment of the present disclosure is an air-conditioning system including an air-conditioning controller and a server. The air-conditioning controller is configured to control an air-conditioning apparatus that performs air conditioning for an air-conditioned space. The server is configured to communicate with the air-conditioning controller. The air-conditioning controller includes an obtainer, an indoor model constructor, and a first communication unit. The obtainer is configured to obtain room shape information indicating a shape of the air-conditioned space and air state information indicating a state of air in the air-conditioned space. The indoor model constructor is configured to construct an indoor model used for CFD simulation, based on the room shape information and the air state information. The first communication unit is configured to send the indoor model to a server. The server includes a coupled analysis unit and a second communication unit. The coupled analysis unit is configured to execute the CFD simulation based on the indoor model to derive a state quantity indicating the state of air at each time point within a predetermined first time range. The coupled analysis unit is configured to derive a deviation of the state quantity at each time point from a preset target state quantity and integrate the deviations at the individual time points within the first time range. The coupled analysis unit is configured to derive an optimal solution of a control parameter of the air-conditioning apparatus such that an objective function including an integrated value obtained by integrating the deviations is minimized. The second communication unit is configured to send the optimal solution to the air-conditioning controller. The air-conditioning controller further includes a control target determiner and a commander. The control target determiner is configured to determine a control target value, which is a target value of the control parameter, based on the optimal solution obtained from the server. The commander is configured to control the air-conditioning apparatus based on the control target value.

Advantageous Effects of Invention

With the use of the air-conditioning controller and the air-conditioning system according to an embodiment of the present disclosure, the obtainer obtains room shape information and air state information, and the indoor model constructor constructs an indoor model based on the room shape information and the air state information. This enables the indoor model constructor to construct the indoor model based on the air state in the air-conditioned space and enables the coupled analysis unit to derive an optimal solution based on this indoor model. The air-conditioning controller can thus perform air-conditioning control based on the optimal solution that is derived from the indoor model obtained based on the air state in the air-conditioned space. Hence, the air-conditioning controller can efficiently change the current air state in the air-conditioned space to the air state desired by a user, thereby making it possible to speedily improve the comfort level of the user.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is a schematic diagram illustrating an example of the configuration of an air-conditioning system including an air-conditioning controller according to Embodiment 1.

FIG. 2 is a block diagram illustrating an example of the configuration of the air-conditioning controller according to Embodiment 1.

FIG. 3 is a flowchart illustrating target area determination processing executed by a target area determiner according to Embodiment 1.

FIG. 4 is a schematic view illustrating an example of a target area determined by the target area determiner according to Embodiment 1.

FIG. 5 is a flowchart illustrating an example of analysis processing executed by a coupled analysis unit according to Embodiment 1.

FIG. 6 is a flowchart illustrating sensitivity deriving processing executed by the coupled analysis unit according to Embodiment 1.

FIG. 7 is a schematic diagram for explaining the direction of processing for deriving a state quantity and the direction of processing for deriving an adjoint variable along the time axis in Embodiment 1.

FIG. 8 is a flowchart illustrating an example of determination processing for a control target value at a specific timing executed by the air-conditioning controller according to Embodiment 1.

FIG. 9 is a diagram illustrating an example of the hardware configuration of the air-conditioning controller according to Embodiment 1.

FIG. 10 is a block diagram illustrating an example of the configuration of a target area determiner according to Embodiment 2.

FIG. 11 is a schematic view illustrating an example of a user stay area based on position information in Embodiment 2.

FIG. 12 is a schematic diagram illustrating an example of a neural network in Embodiment 2.

FIG. 13 is a flowchart illustrating an example of obtaining processing for radio wave information executed by an air-conditioning controller according to Embodiment 2.

FIG. 14 is a flowchart illustrating an example of learning processing executed by a learning model generator according to Embodiment 2.

FIG. 15 is a flowchart illustrating an example of target area determination processing executed by the air-conditioning controller according to Embodiment 2.

FIG. 16 is a diagram for explaining processing for deriving an adjoint variable and sensitivity executed by a coupled analysis unit according to Embodiment 3.

FIG. 17 is a flowchart illustrating an example of sensitivity deriving processing executed by the coupled analysis unit in Embodiment 3.

FIG. 18 is a block diagram illustrating an example of the configuration of an air-conditioning system according to Embodiment 4.

FIG. 19 is a diagram illustrating an example of the hardware configuration of an air-conditioning controller according to Embodiment 4.

FIG. 20 is a diagram illustrating an example of the hardware configuration of a server according to Embodiment 4.

DESCRIPTION OF EMBODIMENTS

Air-conditioning controllers according to embodiments will be described below in detail with reference to the drawings.

Embodiment 1

FIG. 1 is a schematic diagram illustrating an example of the configuration of an air-conditioning system including an air-conditioning controller according to Embodiment 1. An air-conditioning system 100 according to Embodiment 1 includes an air-conditioning apparatus 1, a detector 2, and an air-conditioning controller 3. The air-conditioning controller 3 is connected to the air-conditioning apparatus 1 and the detector 2 via a telecommunication line 5 to communicate with each other. The air-conditioning controller 3 may be a device that performs wireless communication with the air-conditioning apparatus 1 and the detector 2.

The air-conditioning apparatus 1 includes an outdoor unit 10, an indoor unit 11, and a remote controller 12. The outdoor unit 10 and the indoor unit 11 are connected to each other via a heat medium pipe, which is not shown. The outdoor unit 10 cools or heats a heat medium. The indoor unit 11 performs heat exchange between a heat medium and air in an air-conditioned space, thereby controlling the temperature in the air-conditioned space. Examples of the heat medium are brine, water, and refrigerant. Examples of the refrigerant are widely used HFC refrigerant, such as R32, R410A, and R448A, and environmentally friendly natural refrigerant, such as R290 and R717.

The remote controller 12 receives an instruction from a user to the air-conditioning apparatus 1 and causes the outdoor unit 10 and the indoor unit 11 to operate based on this instruction. Examples of the instruction are an operation start instruction, an operation stop instruction, and other instructions, such as those concerning the set temperature, air volume, and airflow direction. Hereinafter, information indicating an instruction input by a user to the remote controller 12 may also be called instruction information.

The detector 2 includes multiple sensors 20 for obtaining various types of information on the air-conditioned space. For example, at least one of the sensors 20 is a thermographic device, which obtains information indicating the temperature distribution in the air-conditioned space. The information indicating the temperature distribution is information of a thermal image, for example. Hereinafter, the information indicating the temperature distribution may also be called temperature distribution information. The temperature distribution information is an example of occupant stay information indicating an area where a user stays within the air-conditioned space.

The sensors 20 include one or more temperature sensors for detecting the temperature within the air-conditioned space. If the detector 2 includes two or more temperature sensors, the temperature sensors are installed at different positions within the air-conditioned space and detect the temperatures at the individual installation positions. Hereinafter, information indicating the temperature in the air-conditioned space may also be called temperature information. The temperature information is an example of air state information indicating the state of air in the air-conditioned space.

The sensors 20 may include one or more flow velocity meters for detecting the flow velocity of air within the air-conditioned space. If the detector 2 includes two or more flow velocity sensors, the flow velocity sensors are installed at different positions within the air-conditioned space and detect the flow velocities of air at the individual installation positions. Hereinafter, information indicating the flow velocity of air in the air-conditioned space may also be called flow velocity information. The flow velocity information is an example of the air state information.

The sensors 20 may include one or more air pressure sensors for detecting the pressure of air in the air-conditioned space. If the detector 2 includes two or more air pressure sensors, the air pressure sensors are installed at different positions within the air-conditioned space and detect pressures of air at the individual installation positions. Hereinafter, information indicating the pressure of air in the air-conditioned space may also be called air pressure information. The air pressure information is an example of the air state information.

The sensors 20 may include at least one of a photoelectric sensor, a camera-built-in laser displacement sensor, an ultrasonic sensor, and an image recognition sensor, for example. This enables the detector 2 to obtain information on the shape and the dimensions of the air-conditioned space, such as the length, width, and height. Hereinafter, information on the shape and the dimensions of the air-conditioned space may also be called room shape information.

The air-conditioning controller 3 controls the air-conditioning apparatus 1 based on various types of information, such as the above-described room shape information, air state information, and instruction information. FIG. 2 is a block diagram illustrating an example of the configuration of the air-conditioning controller according to Embodiment 1. The air-conditioning controller 3 includes an obtainer 30, a storage 31, a current state estimator 32, an indoor model constructor 33, a device model constructor 34, a target area determiner 35, a coupled analysis unit 36, a control target determiner 37, and a commander 38.

The obtainer 30 obtains instruction information and device information from the air-conditioning apparatus 1. The device information is information on the performance of the air-conditioning apparatus 1, and more specifically, information partly indicating the values that can be set as the values of control parameters of the air-conditioning apparatus 1. The device information includes at least one of information indicating the rated capacity and the operating range of a compressor and other components forming a refrigerant circuit provided in the air-conditioning apparatus 1, information indicating the operating range of a fan that sends air to a heat exchanger in the refrigerant circuit, and information indicating the operating range of an airflow direction louver for adjusting the direction of airflow to the air-conditioned space. The operating range of the compressor is a range of frequencies available to the compressor. The operating range of the fan is a range of frequencies available to the fan. The fan is installed in each of the indoor unit 11 and the outdoor unit 10. Information indicating the operating range of the fan in the indoor unit 11 corresponds to information indicating the volume of air sent from the indoor unit 11 to the air-conditioned space. The operating range of the airflow direction louver corresponds to information indicating the direction of air sent from the indoor unit 11 to the air-conditioned space. The obtainer 30 may obtain device information from an external server, for example, based on information for identifying the air-conditioning apparatus 1, such as the production number of the air-conditioning apparatus 1.

The obtainer 30 obtains room shape information from the detector 2 or an external device. Room shape information obtained from an external device may be information created based on building information modeling (BIM), for example. The obtainer 30 obtains temperature distribution information and air state information from the detector 2. The obtainer 30 may include all or some of the sensors 20. That is, all or some of the sensors 20 may be included in the air-conditioning controller 3. The storage 31 stores various types of information, such as the room shape information and device information, obtained by the obtainer 30.

The current state estimator 32 estimates the current state of air within the air-conditioned space, based on the air state information obtained by the obtainer 30. For example, if multiple temperature sensors are installed in the air-conditioned space, the current state estimator 32 obtains the average temperature in the air-conditioned space at a current time point, based on temperature information obtained from each of the temperature sensors. The current state estimator 32 then generates information representing the correlation between the average temperature and information of the current time point. The information of the current time point is a current time, for example. Hereinafter, information of the current time point may also be called current time information. If a single temperature sensor is installed in the air-conditioned space, the current state estimator 32 generates information representing the correlation between the temperature information obtained from this temperature sensor and the current time information.

If multiple flow velocity meters are installed in the air-conditioned space, the current state estimator 32 obtains the average flow velocity of air in the air-conditioned space at the current time point, based on flow velocity information obtained from each of the flow velocity meters. The current state estimator 32 then generates information representing the correlation between the average flow velocity and the current time information. If a single flow velocity meter is installed in the air-conditioned space, the current state estimator 32 generates information representing the correlation between the flow velocity information obtained from this flow velocity meter and the current time information.

If multiple air pressure sensors are installed in the air-conditioned space, the current state estimator 32 obtains the average air pressure in the air-conditioned space at the current time point, based on air pressure information obtained from each of the air pressure sensors. The current state estimator 32 then generates information representing the correlation between the average air pressure and the current time information. If a single air pressure sensor is installed in the air-conditioned space, the current state estimator 32 generates information representing the correlation between the air pressure information obtained from this air pressure sensor and the current time information.

The indoor model constructor 33 constructs an indoor model based on the room shape information obtained by the obtainer 30 and the information generated by the current state estimator 32. The indoor model constructor 33 stores the constructed indoor model in the storage 31. The indoor model is a model concerning an airflow in the air-conditioned space and including information for simulation based on computational fluid dynamic (CFD) conducted by the coupled analysis unit 36, which will be discussed later. Hereinafter, such simulation will be called CFD simulation. In the CFD simulation, the air-conditioned space is divided into a set of multiple small regions in a lattice-like form and is then subjected to simulation processing.

In more detail, the indoor model is a model for solving governing equations for a fluid, which will be discussed later, in the CFD simulation and is a model concerning an airflow that reflects the geometric shape of the air-conditioned space. The indoor model includes information on the state quantity indicating the state of the airflow and information indicating a boundary condition for a state field. The state quantity is the temperature, flow velocity, and pressure of air. The state field is a field in the air-conditioned space and is a field obtained by the state quantity at each of the positions in the air-conditioned space. The boundary condition is a condition concerning the influence of boundaries, such as the walls, floor, and ceiling, on the state field. The indoor model includes information concerning the state quantity and the boundary condition at each time point within a predetermined first time range, which allows for the execution of unsteady simulation.

Based on device information, the device model constructor 34 constructs a device model that associates the coefficient of performance (COP) of the air-conditioning apparatus 1 with control parameters, that is, a device model indicating the COP as a function of the control parameters. The device model constructor 34 stores the constructed device model in the storage 31. The device model is represented by the following expression 1, for example.

[ Math . 1 ]  COP = f RAC ( V inlet , T inlet ) ( 1 )

In expression 1, V_inlet is the outlet velocity, T_inlet is the outlet temperature, and f_RAC is a function. The outlet velocity is the velocity of air blown out of the air outlet of the air-conditioning apparatus 1. The outlet temperature is the temperature of air blown out of the air outlet. The outlet velocity and the outlet temperature are examples of the control parameters.

The device model includes a constraint based on device information. The constraint is a condition for the values available to a control parameter of the air-conditioning apparatus 1. For example, the constraint is a condition for one or both of the upper limit and the lower limit of the values available to a control parameter. The constraint is represented by the following expression 2, for example.

[ Math . 2 ]  { V min ≤ V inlet ≤ V max T min ≤ T inlet ≤ T max ( 2 )

In expression 2, V_min is the lower limit value of the outlet velocity, Vmax is the upper limit value of the outlet velocity, T_min is the lower limit value of the outlet temperature, and T_max is the upper limit value of the outlet temperature. The device model may be a model that includes the constraint represented by expression 2 but does not include the COP as a function represented by expression 1.

The target area determiner 35 determines a target area to be air-conditioned, based on temperature distribution information. FIG. 3 is a flowchart illustrating target area determination processing executed by the target area determiner of Embodiment 1. In step S1, the target area determiner 35 refers to temperature distribution information and identifies a heat source in the air-conditioned area. More specifically, the target area determiner 35 identifies certain factors about a heat source, such as the position, size, and shape of the heat source. In step S2, the target area determiner 35 determines the presence or the absence of occupants in the air-conditioned space, based on the identified factors about the heat source, such as the size and the shape of the heat source, and a temporal change in the position, size, and shape of the heat source, for example. If there is no occupant (NO in step S2), the target area determiner 35 returns to step S1. If there is an occupant (YES in step S2), the target area determiner 35 extracts an area where the occupant is present (hereinafter called the occupant presence area) in step S3. In addition to the occupant presence area, the target area determiner 35 may also extract the surrounding area. The surrounding area is an area within a predetermined distance from the occupant presence area. The predetermined distance is 50 [cm] to 1 [m], for example. In step S4, the target area determiner 35 stores information indicating the area extracted in step S3 in the storage 31. Hereinafter, an area including the position of an occupant extracted by the target area determiner 35 in step S3 may also be called a candidate area, and information indicating a candidate area may also be called candidate area information.

In step S5, the target area determiner 35 determines whether a predetermined determination time has elapsed. If the determination time has not elapsed (NO in step S5), the target area determiner 35 returns to step S1. If the target area determiner 35 returns from step S5 to step S1 and re-executes step S4, it stores in step S4 candidate area information indicating a candidate area extracted in step S3, together with the candidate area information previously stored in step S4. That is, candidate area information is cumulatively stored in the storage 31.

If the determination time has elapsed (YES in step S5), in step S6, the target area determiner 35 obtains information indicating the frequency of the presence of an occupant (hereinafter called the occupant presence frequency) in each of the areas of the air-conditioned space, based on the candidate area information stored in the storage 31. In step S7, the target area determiner 35 determines whether there is a candidate area of which occupant presence frequency is higher than or equal to a predetermined threshold frequency.

If there is no candidate area of which occupant presence frequency is higher than or equal to the predetermined threshold frequency (NO in step S7), the target area determiner 35 returns to step S1. In this case, the target area determiner 35 may restart counting the determination time.

If there is a candidate area of which occupant presence frequency is higher than or equal to the predetermined threshold frequency (YES in step S7), the target area determiner 35 determines this candidate area to be a target area in step S8. The target area is an area to be air-conditioned by the air-conditioning apparatus 1 in the air-conditioned space.

In step S9, the target area determiner 35 stores information indicating the determined target area in the storage 31. At this time, the target area determiner 35 may delete the stored candidate area information from the storage 31.

The target area may be determined based on the content of an operation input into the air-conditioning controller 3 by a user, instead of or in addition to the temperature distribution information. More specifically, the air-conditioning controller 3 may receive from a user a certain setting indicating the area where the user stays within the air-conditioned space or the area desired by the user to be air-conditioned, for example, and determine the target area based on the received setting. The user may input a setting by using one or both of the air-conditioning apparatus 1 and a terminal device of the user. Information indicating the area where the user stays set into the air-conditioning controller 3 is an example of the occupant stay information.

FIG. 4 is a schematic view illustrating an example of a target area determined by the target area determiner according to Embodiment 1. In the example in FIG. 4, the determination time is defined as three days. The thick white arrow in FIG. 4 indicates the direction of time. As shown in FIG. 4, among the multiple sensors 20, the sensor 20 for obtaining temperature distribution information is installed in the indoor unit 11. The target area determiner 35 extracts a candidate area CA including the position of a user H, based on the temperature distribution information obtained by the obtainer 30 from the sensor 20. The temperature distribution information indicates the temperature distribution within the air-conditioned space in each of the first to third days. In FIG. 4, the candidate area CA in each of the first to third days is represented by a conical shape. The apex of this conical shape indicates the position of the sensor 20 for obtaining the temperature distribution information, and the bottom surface of the conical shape is the surface indicated by the hatched portion. The target area determiner 35 determines a target area Ω, based on the candidate area information indicating these candidate areas CA. In FIG. 4, the target area Q is represented by a conical shape of which apex indicates the position of the sensor 20 and of which bottom surface is an elliptical surface indicated by the broken line and by the hatched portion inside. The target area determiner 35 may determine the height of the target area Q from the floor, based on the height or the sitting height of the user, as well as the area in the horizontal direction. In FIG. 4, the height of the target area Q from the floor surface is indicated by the area with the hatched portion outside the elliptical surface indicated by the broken line.

The coupled analysis unit 36 executes optimization processing for control parameters of the air-conditioning apparatus 1, based on the indoor model constructed by the indoor model constructor 33, the target area determined by the target area determiner 35, and the device model constructed by the device model constructor 34. The optimal value of a control parameter is the value when the objective function, which will be discussed later, is minimized. Hereinafter, the optimal value of a control parameter may also be called the optimal solution. The coupled analysis unit 36 derives the optimal solution based on expression 3.

[ Math . 3 ]  minimize u in ( τ ) , θ in ( τ ) ⁢ 𝒥 = 𝒥 ⁡ ( 𝒲 , 𝒰 ) ( 3 ) s . t . ⁢ ℛ ⁡ ( 𝒲 , 𝒰 ) = 0 ,

In expression 3, J is an objective function, T is a time in the first time range, and W is a state variable vector representing the state quantity, such as the flow velocity and the temperature of air, in each of the small regions in the air-conditioned space. Components of the state variable vector W correspond to the above-described state quantity. In expression 3, U is a control variable vector, such as the outlet velocity and the outlet temperature. Components of the control variable vector U correspond to control parameters. In expression 3, U_in(T) is a time series condition for the outlet velocity vector, while θ_in(T) is a time series condition for the outlet temperature, and U_in(T) and θ_in(T) are included in U. Each time series condition of U_in(T) and θ_in(T) satisfies expression 2. In expression 3, “s.t. R(W,U)=0” indicates that satisfying the governing equations is a condition. R(W,U) is a function using the state variable vector W and the control variable vector U that satisfy the governing equations.

The governing equations are represented by the following expressions 4, 5, and 6.

[ Math . 4 ]  ∇ · u = 0   ( 4 ) [ Math . 5 ]  ρ ⁡ ( ∂ u ∂ t + ( u ·   ∇ u ) ) = - ∇ p + ∇ · ( μ ⁢ ∇ u ) + ( ρ - ρ 0 ) ⁢ g ( 5 ) [ Math . 6 ]  ρ ⁢   C p ( ∂ T ∂ t + u · ∇ T ) =   ∇ · ( k ⁢ ∇ T ) + Q   ( 6 )

Expression 4 is a continuity equation representing the mass conservation of a fluid. Expression 5 is an incompressible Navier-Stokes equation representing the momentum conservation. Expression 6 is an energy equation. In expressions 4 to 6, U is the flow velocity of air in each small region. In expressions 5 and 6, ρ is the density of air, and t is the time. In expression 5, p is the pressure vector of air in each small region, μ is the viscosity of air in each small region, and ρ₀ is the reference density of air. The third term of the right-hand side of expression 5 represents the buoyancy, in which g is the value representing the magnitude of an acceleration vector, such as a gravity vector. In expression 6, C_p is the specific heat at constant pressure, T is the temperature of air in each small region, k is thermal conductivity, and Q is the amount of heat supplied to each small region. If necessary, preset values may be used for some of the above-described values.

J in expression 3 is represented by expression 7.

[ Math . 7 ]  J opt   = J 1   + ω × J 2 ( 7 )

J_opt in expression 7 corresponds to J in the above-described expression 3. J₁ in expression 7 is represented by expression 8, and J₂ is represented by expression 9.

[ Math . 8 ]  J 1 = ∫ Ω ∫ 0 T [ γ T ( θ ⁡ ( x , y , z , τ ) - θ d ) 2 + γ v ( u ⁡ ( x , y , z , τ ) - u d ) 2 ] ⁢ dV ⁢ d ⁢ τ ( 8 ) [ Math . 9 ]  J 2 = Q RAC COP ( 9 )

In expression 7, ω is a preset coefficient, which may be set based on a setting input by a user and the operating situation of the air-conditioning apparatus 1, for example. In expression 7, ω may be a vector or a matrix including multiple coefficients.

In expression 8, x, y, and z are the x coordinate, y coordinate, and z coordinate, respectively, in the air-conditioned space. In expression 8, T is the time in the first time range, θ(x, y, z, T) is the temperature of air at time T at a position in the air-conditioned space of which coordinates are represented by (x, y, z), and θ_d is the target temperature of air, which is set to improve the degree of comfort of a user. In expression 8, U(x, y, z, T) is the flow velocity of air at time T at a position in the air-conditioned space of which coordinates are represented by (x, y, z), and U_d is the target flow velocity of air, which is set to improve the degree of comfort of a user. Hereinafter, θ_d and U_d may be called the target state quantity. The target state quantity is based on instruction information input into the remote controller 12 by a user and obtained by the air-conditioning controller 3. The target state quantity is stored in the storage 31.

In expression 8, Ω is the target area, and T is the length of the first time range. In expression 8, the start time point and the end time point of the first time range is set to 0 and T, respectively. The length of the first time range may be a preset fixed value or may be a variable value that can be changed by a user. In expression 8, γT is a weight coefficient applied to the term concerning the temperature of air, and γ_v is a weight coefficient applied to the term concerning the flow velocity of air. γT and γ_v may be preset values. Alternatively, each weight coefficient of γT and γ_v may be set based on at least one of the setting set by a user and the operating situation of the air-conditioning apparatus 1, or may be a vector or a matrix including multiple coefficients, for example.

As represented in expression 8, J₁ is represented by a deviation of the state quantity in the target area at each time point within the first time range from the target state quantity. In expression 9, Q_RAC is the air-conditioning capacity and is represented by expression 10.

[ Math . 10 ]  Q RAC = ∫ 0 τ ∑ i n ρ ⁢ CV i ⁢ A i ( T i , inlet - T i , outlet ) ( 10 )

In expression 10, T in ‘j’ that represents the integral, that is, T representing the upper limit of the integral, is the length of the first time range. In expression 10, ρ is the density of air, C is the specific heat at constant pressure. In expression 10, V_i is the outlet velocity of the i-th air conditioning apparatus 1, and n air-conditioning apparatuses 1 are defined as being disposed in the air-conditioned space. In expression 10, A_i is the surface area of the air outlet of the i-th air-conditioning apparatus 1, T_i,inlet is the outlet temperature of the i-th air conditioning apparatus 1, and T_i,outlet is the temperature of air sucked into the i-th air-conditioning apparatus 1.

The coupled analysis unit 36 determines the control variable vector U in expression 3 such that expressions 4, 5, and 6 are satisfied and such that the objective function indicated by J_opt in expression 8 is minimized. If the objective function is minimized, one or both of the following expressions 11 and 12 hold true.

[ Math . 11 ]  ❘ "\[LeftBracketingBar]" J k + 1 → J k ❘ "\[RightBracketingBar]" < ε 1 ( 11 ) [ Math . 12 ]  ❘ "\[LeftBracketingBar]" ∇ F k ❘ "\[RightBracketingBar]" < ε 2 ( 12 )

Expressions 11 and 12 represent the convergence of the objective function. In expressions 11 and 12, k is the number of times the objective function is updated by the coupled analysis unit 36, and ε₁ and ε₂ are desired small quantities. The left-hand side in expression 12 corresponds to the sensitivity, which will be discussed later. The coupled analysis unit 36 updates U in expression 3 until one or both of expressions 11 and 12 are satisfied.

Analysis processing executed by the coupled analysis unit 36 will be described below in detail with reference to FIG. 5. FIG. 5 is a flowchart illustrating an example of analysis processing executed by the coupled analysis unit according to Embodiment 1. The coupled analysis unit 36 repeats steps S11 through S15 until the value of the objective function is minimized.

In step S11, the coupled analysis unit 36 performs CFD simulation based on the indoor model. The coupled analysis unit 36 conducts forward analysis as CFD simulation. The forward analysis is analysis to acquire the state quantity in each of the small regions forming the air-conditioned space from the boundary conditions and governing equations, for example. That is, based on the indoor model, the coupled analysis unit 36 solves the governing equations represented by expressions 4, 5, and 6 to derive the state quantity in each small region. When conducting forward analysis, to derive the state field that changes over time due to certain conditions, such as the boundary conditions, the coupled analysis unit 36 derives the state quantity in the forward direction of time.

In the indoor model, the boundary conditions at the air outlet of the air-conditioning apparatus 1 are indicated by U_in(T) and θ_in(T) in expression 3. Hereinafter, components of the control variable vector U including U_in(T) and θ_in(T), which are boundary conditions, at each time point within the first time range may also be called a first value. The first value is set to be a predetermined initial value in step S11, which is a step before step S15 is executed. Based on the device model, the coupled analysis unit 36 determines the first value for each time point within the first time range. The coupled analysis unit 36 may determine the first value based on a setting set by a user, such as the set temperature or the set air velocity, for example.

In step S12, the coupled analysis unit 36 derives the objective function, based on the state quantity derived for each small region in step S11, the control variable vector U including U_in(T) and θ_in(T), and the target area Ω determined by the target area determiner 35.

In step S13, the coupled analysis unit 36 derives the sensitivity. The sensitivity corresponds to the degree of influence on the objective function when the value of a control parameter is changed. The sensitivity is represented by expression 13.

[ Math . 13 ]  ∂ ℒ ∂ v in , y = p α , in - 1 Re ⁢ ( n i ∂ ∂ x i ) ⁢ v in - λθ in , ( 13 ) ∂ ℒ ∂ θ in = 1 Pe ⁢ ( n i ∂ ∂ x i ) ⁢ T a , in - λ ⁢ u in

Expression 13 represents the dimensionless sensitivity. L is Lagrangian, and the sensitivity is expressed by a first variation of L. L is defined by the following expression 14 that represents an unconstrained problem based on the Lagrangian relaxation method to solve the optimization problem represented by expression 3.

[ Math . 14 ]  minimize ⁢ L = J + 〈 P , R 〉 ( 14 )

J, which is the first term of the right-hand side in expression 14, is the objective function. <P,R> in the second term is the inner product of P and R, where P is an adjoint variable vector and R is R in expression 3. As a result of minimizing L obtained by adding the inner product of R and P to J when R in expression 3 is 0, the minimized J is obtained.

The subscript i in x_i in expression 13 is one of 1, 2, and 3, and x₁ is the x coordinate in the air-conditioned space, and x₂ is the y coordinate in the air-conditioned space, and x₃ is the z coordinate in the air-conditioned space. In expression 13, p_a,in is the adjoint variable of the air pressure at the air outlet, T_a,in is the adjoint variable of the air temperature at the air outlet, and v_in is the adjoint variable of the flow velocity of air at the air outlet. The adjoint variable is a variable used as an actual physical quantity to derive the sensitivity. Details of the adjoint variable will be discussed later.

In expression 13, n_i is a component of a vector normal to the air outlet. In expression 13, it is defined that the blowing direction of air from the air-conditioning apparatus 1 is the y direction. In expression 13, v_in,y is a y component of the outlet flow velocity, θ_in is the outlet temperature, and U_in is a component of the outlet velocity. In expression 13, it is defined that θ_in and U_in are dimensionless values.

λ in expression 13 is represented by expression 15.

[ Math . 15 ]  λ = [ p a , in - ( 1 / Re ) ⁢ { n i ( ∂ / ⁢ ∂ x i ) } ⁢ v in ] / θ in ( 15 )

In expression 15, p_a,in, n_i, x_i, v_in, and θ_in are similar to those in expression 13. Re in expressions 13 and 15 is a dimensionless number and is represented by expression 16.

[ Math . 16 ]  Re = V ref ⁢ L ref / μ ( 16 )

In expression 16, μ is a coefficient of kinematic viscosity, V_ref is a reference velocity, and L_ref is a reference length. The reference velocity and the reference length are given values dependent on the situation of a target flow field. The reference velocity is the outlet velocity, for example, while the reference length is the length of the width of the air outlet, for example. The flow field is a vector field in which the flow velocity is defined at a given point in a space.

Pe in expression 13 is a dimensionless number and is represented by expression 17.

[ Math . 17 ]  Pe = V ref ⁢ L ref / κ ( 17 )

V_ref and L_ref in expression 17 are similar to those in expression 16. In expression 17, κ is the temperature diffusivity.

FIG. 6 is a flowchart illustrating sensitivity deriving processing executed by the coupled analysis unit according to Embodiment 1. Steps S21 and S22 in FIG. 6 correspond to step S13 in FIG. 5. In step S21, the coupled analysis unit 36 derives the value of the adjoint variable from the state quantity derived in step S11, based on the following expressions 18, 19, and 20.

[ Math . 18 ]  ∂ v j / ∂ x j = 0 ( 18 ) [ Math . 19 ]  - ∂ v i / ∂ τ + v j ( ∂   u j / ∂ x i ) - u j ⁢ ( ∂ v i / ∂ x j ) + T a ⁢ ( ∂ θ / ∂ x i ) - ∂ / ⁢ ∂ x j ⁢ { ( 1 / Re ) ⁢ ∂ v i / ∂ x j } + ∂ p a / ∂ x i + α ⁢ ( u i - u d ) = 0 ( 19 ) [ Math . 20 ]  - ∂ T a / ∂ τ - Ri ⁢ δ i , 2 ⁢ v i - u j ( ∂ T a / ∂ x j ) - ∂ / ⁢ ∂ x j ⁢ { ( 1 / Pe ) ⁢ ∂ T a / ∂ x j } + β ⁢ θ = 
 0 ( 20 )

The subscripts i and j in expressions 18, 19, and 20 each indicate the x coordinate, y coordinate, and z coordinate in the air-conditioned space. That is, each subscript of i and j is one of 1, 2, and 3, and x₁ is the x coordinate in the air-conditioned space, and x₂ is the y coordinate in the air-conditioned space, and x₃ is the z coordinate in the air-conditioned space. The symbols in expressions 18, 19, and 20 are as follows: p_a in expression 19 is p_a,in in expression 13; T_a in expression 20 is T_a,in in expression 13; v_i in expressions 18, 19, and 20 is an i component of v_in in expression 13; v_j in expressions 18, 19, and 20 is a j component of v_in in expression 13; 0 in expressions 19 and 20 is θ_in in expression 13; U_i in expressions 19 and 20 is an i component of U_in in expression 13; and U_j in expressions 19 and 20 is a j component of U_in in expression 13. Re and Pe are similar to those described above. Regarding δ_i,2, when the subscript i is 2, δ_i,2 is 1, and when the subscript i is other than 2, δ_i,2 is 0. The direction of i=2 is the direction of a line normal to the air outlet, that is, the blowing direction of air, which is the y direction, as described above.

In expressions 19 and 20, α and β are coefficients. In Embodiment 1, it is defined that α and β are greater than 0 inside the target area Ω and are 0 outside the target area Ω. Ri in expression 20 is represented by the following expression 21.

[ Math . 21 ]  Ri = [ g ⁢ β 00 ⁢ Δ ⁢ T ref ⁢ L ref ] / V ref 2 ( 21 )

g in expression 21 is similar to g in expression 5. V_ref and L_ref in expression 21 are similar to those discussed above. In expression 21, β₀₀ is the coefficient of cubical expansion, and T_ref is the reference temperature. The reference temperature is a given value dependent on the situation of a target flow field and is the outlet temperature, for example.

In step S21, the coupled analysis unit 36 conducts inverse analysis to derive the adjoint variable. Inverse analysis is analysis conducted by reversing the cause and effect relationship, that is, the input and output relationship, of forward analysis, and is a technique for deriving boundary conditions, for example, used for obtaining the state quantity, which is the output result of forward analysis, from this state quantity. To derive boundary conditions, for example, from the state field, which has changed over time due to the boundary conditions, the coupled analysis unit 36 derives the boundary conditions along the backward direction of time. That is, in Embodiment 1, the adjoint variable corresponding to the boundary conditions at the air outlet is derived along the backward direction of time.

FIG. 7 is a schematic diagram for explaining the direction of processing for deriving the state quantity and the direction of processing for deriving the adjoint variable along the time axis in Embodiment 1. As shown in FIG. 7, in step S11 in FIG. 5, based on the time point t in the first time range, the coupled analysis unit 36 sets the start time point to be t₀ and the end time point to be (t₀+T) and sequentially derives the state quantity from t₀ to (t₀+T). T is the length of the first time range. The coupled analysis unit 36 stores the derived state quantity at each time point t from t₀ to (t₀+T) in the storage 31.

As illustrated in FIG. 7, in step S21 in FIG. 6, based on expressions 18, 19, and 20, the coupled analysis unit 36 derives the adjoint variable at each time point t from the state quantity at the corresponding time point t stored in the storage 31. In step S21, the coupled analysis unit 36 sets the start time point to be (t₀+T) and the end time point to t₀ and sequentially derives the adjoint variable from (t₀+T) to t₀ along the backward direction of time.

In step S22 in FIG. 6, based on expression 13, the coupled analysis unit 36 derives the sensitivity from the state quantity derived in step S11 and the adjoint variable derived in step S21.

In step S14 in FIG. 5, the coupled analysis unit 36 updates the control variable vector based on the following expression 22.

[ Math . 22 ]  x ⁡ ( t ) k + 1 = x ⁡ ( t ) k + α ⁡ ( t ) k ⁢ d k ( 22 )

In expression 22, k is the number of times the control variable vector is updated, x(t)^k is the control variable vector that has not been updated, and x(t)^k+1 is the updated control variable vector. That is, x(t)^k is the control variable vector obtained by the k-th updating time, and x(t)^k+1 is the control variable vector obtained by the (k+1)-th updating time. In expression 22, α(t)^k is a sensitivity vector when the updating time is k and includes a component related to the sensitivity of each of the control parameters, such as the outlet temperature and the outlet velocity. In expression 22, d^k is the updating amount of each component of the control variable vector. Expression 22 may be an expression based on the update technique using the method of steepest descend or may be, for example, an expression based on the quasi-Newton method using sensitivity information.

In step S15, the coupled analysis unit 36 determines whether the objective function converges. That is, the coupled analysis unit 36 determines whether expression 11 holds true by the objective function obtained from the control variable vector that has not been updated and the objective function obtained from the updated control variable vector. Alternatively, the coupled analysis unit 36 may determine whether the objective function converges by determining whether the sensitivity derived in step S13 satisfies expression 12. In this case, step S15 may be executed after step S13 and before step S14 or in parallel with step S14. Expression 12 may be modified, for example, into the following expression 23. L, v_in,y, and θ_in in expression 23 are similar to those in expression 13. In expression 23, δ is a given small quantity.

[ Math . 23 ]  [ ( 1 / 2 ) · { ( ∂ L / ∂ v in , y ) 2 + ( ∂ L / ∂ θ in ) 2 } ] 1 / 2 < δ ( 23 )

If it is determined in step S15 that the objective function does not converge (NO in step S15), the coupled analysis unit 36 returns to step S11. In this case, after returning from step S15, the coupled analysis unit 36 executes step S11 and the subsequent steps, based on the updated control variable vector. If it is determined in step S15 that the objective function converges (YES in step S15), the coupled analysis unit 36 completes the analysis processing.

Before or after step S15 or in parallel with step S15, the coupled analysis unit 36 stores the control variable vector obtained in step S14 in the storage 31. The control variable vector stored in the storage 31 when it is determined that the objective function converges in step S15 is a vector having the optimal solution of the control parameter as a component. The coupled analysis unit 36 stores the control variable vector obtained in step S14 in association with the indoor model and the device model. The coupled analysis unit 36 uses the condition represented by the control variable vector obtained in step S14 for the boundary condition concerning the air outlet of the air-conditioning apparatus 1 among the boundary conditions of the indoor model. When re-executing step S11 and the subsequent steps, the coupled analysis unit 36 updates the above-described first value based on the device model by using the values of the components of the control variable vector obtained in step S14.

Reference is back to FIG. 2. The control target determiner 37 in Embodiment 1 determines the temperature at a first position to be the control target temperature, based on the temperature distribution within the air-conditioned space achieved by the optimal solution obtained by the coupled analysis unit 36. As the first position, the installation position of the above-described temperature sensor or a position within the target area, for example, may be used. The control target determiner 37 stores the determined control target temperature in the storage 31 in association with the indoor model, the device model, and the optimal solution. The control target determiner 37 may determine the flow velocity of air at a second position to be the control target velocity and store it in the storage 31, based on the flow velocity distribution of air within the air-conditioned space achieved by the optimal solution obtained by the coupled analysis unit 36. As the second position, the installation position of the above-described flow velocity meter or a position within the target area, for example, may be used. The control target determiner 37 may determine the air pressure at a third position to be the control target pressure and store it in the storage 31, based on the pressure distribution of air within the air-conditioned space achieved by the optimal solution obtained by the coupled analysis unit 36. As the third position, the installation position of the above-described air pressure meter or a position within the target area, for example, may be used. Hereinafter, the value of each of the control target temperature, control target velocity, and control target pressure may also be called the control target value.

The commander 38 controls the air-conditioning apparatus 1, based on the control target value derived by the control target determiner 37.

The air-conditioning controller 3 may control the air-conditioning apparatus 1, based on the control target value determined by the control target determiner 37 and stored in the storage 31. However, the indoor model, the device model, and the above-described target state quantity are variable due to changes of the arrangement of furniture within the air-conditioned space, seasonal changes, or changes of settings set by a user, for example. Hence, the air-conditioning controller 3 may reconstruct one or both of the indoor model and the device model at a specific timing, derives the optimal solution based on the reconstructed models, and redetermine the control target value. Examples of the specific timing include a specific time during the day, a specific date and time in one to four weeks, a specific date and time in one to three months, and the start time of the air-conditioning apparatus 1 or the air-conditioning controller 3. The specific timing may be set by a user. Processing executed by the air-conditioning controller 3 at this specific timing will be described below with reference to FIG. 8.

FIG. 8 is a flowchart illustrating an example of determination processing for the control target value at a specific timing executed by the air-conditioning controller according to Embodiment 1. It is defined that, before step S31, the indoor model is constructed by the indoor model constructor 33 and the device model is constructed by the device model constructor 34. In step S31, the control target determiner 37 determines whether a first condition is satisfied. The first condition includes the condition that the indoor model constructed by the indoor model constructor 33 is the same as that stored in the storage 31 and the device model constructed by the device model constructor 34 is the same as that stored in the storage 31. The first condition also includes the condition that the value of the control parameter, such as the outlet temperature and the outlet velocity, represented by the boundary condition regarding the air outlet of the air-conditioning apparatus 1 among the boundary conditions of the indoor model is equal to that of the control parameter that is already stored as the optimal solution. The first condition also includes the condition that the target state quantity, such as the target temperature and the target flow velocity, indicated by newly obtained instruction information is the same as the target state quantity that is already set and stored. Alternatively, the first condition may include at least one of the condition that the indoor model constructed by the indoor model constructor 33 is the same as that stored in the storage 31, the condition that the device model constructed by the device model constructor 34 is the same as that stored in the storage 31, and the condition that newly obtained target state quantity is the same as that stored in the storage 31.

If the first condition is satisfied in step S31 (YES in step S31), the air-conditioning controller 3 proceeds to step S33. If the first condition is not satisfied in step S31 (NO in step S31), in step S32, the coupled analysis unit 36 executes the above-described analysis processing illustrated in FIGS. 5 and 6. Then, the coupled analysis unit 36 stores the obtained optimal solution. After step S32, the air-conditioning controller 3 proceeds to step S33. In step S33, the control target determiner 37 determines the control target value based on the optimal solution stored in the storage 31. In step S34, the commander 38 controls the air-conditioning apparatus 1 based on the control target value determined in step S33.

With the above-described processing, air conditioning in the air-conditioned space can be optimized at an appropriate timing. Before step S31 described above, the target area determiner 35 may redetermine the target area.

The hardware configuration of the air-conditioning controller 3 will be discussed below with reference to FIG. 9. FIG. 9 is a diagram illustrating an example of the hardware configuration of the air-conditioning controller according to Embodiment 1. The air-conditioning controller 3 can be formed by, for example, a first processor 41, a first memory 42, a storage device 43, an input interface circuit 44, and an input-output interface circuit 45 connected to each other via a first bus 40. The first processor 41 is a central processing unit (CPU) or a micro processing unit (MPU), for example. The first memory 42 is a read only memory (ROM) or a random access memory (RAM), for example. The storage device 43 may be a magnetic disk, such as a hard disk drive (HDD), or an optical disc, such as a compact disc (CD). The storage device 43 may be a universal serial bus (USB) memory or a flash memory, such as an SD memory card.

The functions of the obtainer 30 can be implemented by the input interface circuit 44 connected to the multiple sensors 20 and the input-output interface circuit 45 connected to the air-conditioning apparatus 1. The functions of the obtainer 30 may alternatively be implemented by the input interface circuit 44, the input-output interface circuit 45, and the first processor 41. The functions of the storage 31 can be implemented by the storage device 43. The functions of the storage 31 may alternatively be implemented by the first memory 42, in which case, the provision of the storage device 43 in the air-conditioning controller 3 may be omitted. The functions of the current state estimator 32, the indoor model constructor 33, the device model constructor 34, the target area determiner 35, the coupled analysis unit 36, and the control target determiner 37 can be implemented as a result of the first processor 41 reading and executing various programs stored in the first memory 42. The functions of the commander 38 can be implemented by the input-output interface circuit 45. The functions of the commander 38 may alternatively be implemented by the input-output interface circuit 45 and the first processor 41. All or some of the above-described functions of the air-conditioning controller 3 may be implemented by dedicated hardware.

Advantages achieved by the air-conditioning controller 3 according to Embodiment 1 will be described below. The air-conditioning controller 3 according to Embodiment 1 controls the air-conditioning apparatus 1, which performs air conditioning for an air-conditioned space. The air-conditioning controller 3 includes the obtainer 30, the indoor model constructor 33, the coupled analysis unit 36, the control target determiner 37, and the commander 38. The obtainer 30 obtains room shape information indicating the shape of the air-conditioned space and air state information indicating the state of air in the air-conditioned space. The indoor model constructor 33 constructs an indoor model used for CFD simulation, based on the room shape information and the air state information. The coupled analysis unit 36 executes the CFD simulation based on the indoor model. That is, the coupled analysis unit 36 derives a state quantity indicating the state of air at each time point within a predetermined first time range, determines a deviation of the state quantity at each time point from a preset target state quantity, and integrates the deviations at the individual time points within the first time range. Then, the coupled analysis unit 36 derives an optimal solution of a control parameter of the air-conditioning apparatus 1 such that an objective function including the integrated value obtained by integrating the deviations is minimized. The control target determiner 37 determines a control target value, which is a target value of the control parameter, based on the optimal solution. The commander 38 controls the air-conditioning apparatus 1 based on the control target value.

With the above-described configuration, the obtainer 30 obtains room shape information and air state information, and the indoor model constructor 33 constructs the indoor model based on the room shape information and the air state information. The indoor model can thus be constructed based on the current air state in the air-conditioned space. The coupled analysis unit 36 obtains the optimal solution based on such an indoor model, and the air-conditioning controller 3 controls the air-conditioning apparatus 1 based on the optimal solution. The air-conditioning apparatus 1 can thus change the current air state in the air-conditioned space to the air state desired by a user. Hence, even in a state in which the optimal solution has not been obtained in advance, the air-conditioning apparatus 1 can perform air conditioning as desired by a user, thereby making it possible to speedily improve the comfort level of the user.

The obtainer 30 in Embodiment 1 obtains occupant stay information indicating an area where an occupant stays within the air-conditioned space. The air-conditioning controller 3 further includes a target area determiner 35. The target area determiner 35 determines a target area, which is an area to be air-conditioned within the air-conditioned space, based on the occupant stay information. The coupled analysis unit 36 derives the objective function by integrating the deviations at the individual time points within the first time range in the target area. This eliminates the need for the coupled analysis unit 36 to execute integration processing on the entire air-conditioned space to derive the objective function, thereby reducing the processing amount and therefore shortening the time before air conditioning desired by a user is obtained. The air-conditioning controller 3 can thus speedily improve the comfort level of the user.

The occupant stay information in Embodiment 1 is temperature distribution information indicating the temperature distribution in the air-conditioned space. The target area determiner 35 specifies an area where an occupant is present in the air-conditioned space based on the temperature distribution information and obtains the frequency of the presence of the occupant in the specified area. The target area determiner 35 then determines the target area based on the obtained frequency. With this configuration, the air-conditioning controller 3 can set the area where an occupant is present with high frequency to be the area to be air-conditioned, thereby making it possible to improve the comfort level of a user and also to enhance energy saving.

The control target determiner 37 in Embodiment 1 determines the control target value, based on the temperature at a predetermined first position on the temperature distribution in the air-conditioned space achieved by the optimal solution. This can speed up and facilitate processing for determining the control target value, thereby speedily implementing air conditioning desired by a user.

The obtainer 30 in Embodiment 1 obtains device information concerning the performance of the air-conditioning apparatus 1 from the air-conditioning apparatus 1. The air-conditioning controller 3 further includes a device model constructor 34. The device model constructor 34 constructs a device model, which determines a condition for the value of the control parameter, based on the device information. The coupled analysis unit 36 derives the objective function based on the device model. This makes it possible to obtain the optimal solution of the control parameter that reflects the performance of the air-conditioning apparatus 1. The air-conditioning controller 3 can thus suitably control the air-conditioning apparatus 1 according to the performance of the air-conditioning apparatus 1.

The indoor model in Embodiment 1 includes the state quantity at the outlet of air output from the air-conditioning apparatus 1, which is based on the device model. With this configuration, certain factors regarding air output from the installed air-conditioning apparatus 1, such as the outlet velocity and the outlet temperature of air, can be reflected in the boundary conditions of the indoor model, thereby improving the accuracy of the indoor model.

The device model in Embodiment 1 includes information indicating the correlation between the coefficient of performance of the air-conditioning apparatus 1 and the value of the control parameter. This makes it possible to determine the optimal solution that can improve the coefficient of performance.

The obtainer 30 in Embodiment 1 obtains instruction information indicating a setting input into the air-conditioning apparatus by a user. The target state quantity is determined based on the instruction information. With this configuration, the coupled analysis unit 36 can determine the optimal solution that implements air conditioning desired by a user.

The air-conditioning controller 3 further includes a storage 31. The storage 31 stores the indoor model, target state quantity, device model, and optimal solution in association with each other. The control target determiner 37 derives the control target value based on the optimal solution stored in the storage 31 if the target state quantity based on the instruction information that is re-obtained by the obtainer 30 at a specific timing is equal to the target state quantity stored in the storage 31 and if the indoor model based on the air state information that is re-obtained by the obtainer 30 at the specific timing is equal to the indoor model stored in the storage 31 and if the device model based on the device information that is re-obtained by the obtainer 30 at the specific timing is equal to the device model stored in the storage 31. With this configuration, in a state in which there is no change in the air state in the air-conditioned space and in the instruction information and there is also no change in the air-conditioning apparatus 1, the air-conditioning controller 3 performs control based on the stored optimal solution. The air-conditioning controller 3 can thus cause the air-conditioning apparatus 1 to perform air conditioning desired by a user while reducing processing, such as processing for deriving the optimal solution. Hence, the air-conditioning controller 3 can promptly execute control processing, thereby making it possible to more speedily improve the comfort level of a user.

Embodiment 2

The air-conditioning controller 3 according to Embodiment 2 will be described below in detail. The elements of Embodiment 2 similar to those in Embodiment 1 will be designated by the same reference signs. An explanation of the properties such as configurations and functions in Embodiment 2 similar to those in Embodiment 1 will be omitted unless there are special circumstances.

The configuration of the air-conditioning system 100 according to Embodiment 2 is illustrated in FIG. 1 by way of example, as in Embodiment 1. The configuration of the air-conditioning controller 3 according to Embodiment 2 is illustrated in FIG. 2 by way of example, as in Embodiment 1. The hardware configuration of the air-conditioning controller 3 according to Embodiment 2 is illustrated in FIG. 9 by way of example, as in Embodiment 1.

The occupant stay information in Embodiment 2 is temperature distribution information. The obtainer 30 in Embodiment 2 obtains radio wave information on the radio wave intensity and the radio wave receiving direction of a signal indicating an instruction input by a user into the remote controller 12 or the above-described terminal device. The radio wave intensity is the intensity of a radio wave received by the air-conditioning controller 3 or the indoor unit 11 from the remote controller 12 or the terminal device. The radio wave receiving direction is the direction of a radio wave received by the air-conditioning controller 3 or the indoor unit 11 from the remote controller 12 or the terminal device.

Having obtained radio wave information at a certain time point, the obtainer 30 stores in the storage 31 temperature distribution information obtained before or after this time point within a prescribed time period. The obtainer 30 stores the temperature distribution information in association with the radio wave information. The prescribed time period is ten [seconds] to three [minutes], for example. Alternatively, the obtainer 30 may store in the storage 31 the temperature distribution information that is first obtained after the time point at which the radio wave information is obtained or the temperature distribution information that is last obtained before the time point at which the radio wave information is obtained. In this case, too, the obtainer 30 stores the temperature distribution information in association with the radio wave information.

The obtainer 30 may constantly obtain the temperature distribution information from the sensor 20 used for sensing the temperature distribution in the air-conditioned space. Alternatively, upon obtaining radio wave information, the obtainer 30 may send a request signal to this sensor 20 to request it to send temperature distribution information to the air-conditioning controller 3 and obtain temperature distribution information from the sensor 20 as a response to the request signal. If the obtainer 30 includes this sensor 20, the obtainer 30 may obtain temperature distribution information upon obtaining radio wave information or may constantly obtain temperature distribution information.

The obtainer 30 may obtain radio wave information and temperature distribution information multiple times during the above-described determination time. Then, the obtainer 30 may store radio wave information obtained at each time point during the determination time in the storage 31 in association with the temperature distribution information obtained before or after this time point within the prescribed time period. The length of the determination time in Embodiment 2 may be the same as that in Embodiment 1 or may be different from that in Embodiment 1.

The target area determiner 35 in Embodiment 2 is equipped with an artificial intelligence (AI) function. FIG. 10 is a block diagram illustrating an example of the configuration of the target area determiner according to Embodiment 2. The target area determiner 35 in Embodiment 2 includes a learning model generator 350 and an estimator 351.

The learning model generator 350 obtains temperature distribution information and radio wave information in association with each other from the storage 31. Based on the radio wave information, the learning model generator 350 obtains position information indicating the position of the remote controller 12 or the terminal device. The position of the remote controller 12 or the terminal device is estimated to coincide with the position of a user operating the remote controller 12 or the terminal device or to be within 1 [m], for example, from the position of the user. The position information thus corresponds to information indicating the position of the user at a time point at which the obtainer 30 has obtained radio wave information. The learning model generator 350 may obtain the position information based on room shape information, together with the radio wave information. The spatial coordinates of the position information, those of the room shape information, and those of the temperature distribution information are correlated to each other.

The learning model generator 350 conducts machine learning based on a set of the position information and temperature distribution information or a set of the position information, temperature distribution information, and room shape information and generates a learned model. Hereinafter, a combination of the position information and temperature distribution information or a combination of the position information, temperature distribution information, and room shape information may also be called learning data. The learned model may also be called a learning model. In more detail, the learning model generator 350 generates a learning model for estimating a user stay area from the temperature distribution information by using the position information of the learning data as a truth. The learning model generator 350 stores the generated learning model in the storage 31.

FIG. 11 is a schematic view illustrating an example of the user stay area based on position information in Embodiment 2. In FIG. 11, the region indicated by the hatched portion around the user H operating the remote controller 12 is obtained as the user stay area BA. In FIG. 11, the remote controller 12 is located near the user H, and the user stay area BA has been obtained based on the position information indicating the position of the remote controller 12. However, instead of or in addition to the position information of the remote controller 12, the user stay area BA may be obtained based on position information indicating the position of the terminal device of the user H.

The learning model generator 350 generates a learning model based on the algorithm of a known learning method, such as supervised learning, unsupervised learning, and reinforcement learning. An explanation will be given to a case in which the learning model generator 350 generates a learning model based on the algorithm of supervised learning by using a neural network by way of example.

FIG. 12 is a schematic diagram illustrating an example of a neural network in Embodiment 2. The neural network is formed by input layers including multiple neurons, one or more intermediate layers including one or more neurons, and output layers including multiple neurons. In FIG. 12, three input layers X1 through X3 including three neurons, two intermediate layers Y1 and Y2 including two neurons, and three output layers Z1 through Z3 including three neurons are shown.

In FIG. 12, when multiple input values are input into the input layers X1 through X3, intermediate values obtained by applying weights w11 through w16 to the input values are input into the intermediate layers Y1 and Y2. Then, output values obtained by applying weights w21 through w26 to the intermediate values are obtained from the output layers Z1 through Z3. The output values from the output layers Z1 through Z3 are variable based on the values of the weights w11 through w16 and w21 through w26 even if the input values are the same.

The learning model generator 350 inputs temperature distribution information and room shape information into the input layers and conducts learning by adjusting weights, such as the weights w11 through w16 and w21 through w26 shown in FIG. 12, such that the output results from the output layers match or approach the position information, thereby generating a learning model.

Based on the learning model stored in the storage 31, the estimator 351 estimates the position of the user from certain information obtained by the obtainer 30, such as the temperature distribution information, and then determines a target area.

One or both of the learning model generator 350 and the estimator 351 may be disposed outside the air-conditioning controller 3. More specifically, one or both of the learning model generator 350 and the estimator 351 may be disposed in another device, such as a cloud server, that communicates with the air-conditioning controller 3.

More specifically, if the learning model generator 350 is disposed outside the air-conditioning controller 3, the learning model generator 350 receives radio wave information, temperature distribution information, and room shape information obtained by the obtainer 30 from the air-conditioning controller 3. The learning model generator 350 then obtains position information from the received radio wave information and generates a learning model from the position information, temperature distribution information, and room shape information. The learning model generator 350 then sends the generated learning model to the air-conditioning controller 3, and the storage 31 stores the learning model received from the learning model generator 350.

FIG. 13 is a flowchart illustrating an example of obtaining processing for radio wave information executed by the air-conditioning controller according to Embodiment 2. In step S41, the obtainer 30 obtains radio wave information. In step S42, the obtainer 30 stores the radio wave information obtained in step S41 in the storage 31 in association with the following temperature distribution information. The temperature distribution information is information that is obtained within a prescribed time period after the time point at which the obtainer 30 obtains the radio wave information in step S41. Alternatively, the temperature distribution information is information that is obtained within the prescribed time period before the time point at which the obtainer 30 obtains the radio wave information in step S41. Alternatively, the temperature distribution information is information that is obtained for the last time before the time point at which the obtainer 30 obtains the radio wave information in step S41 or information that is obtained for the first time after the time point at which the obtainer 30 obtains the radio wave information in step S41. The obtainer 30 may obtain room shape information together with the temperature distribution information and store the room shape information in the storage 31 in association with the radio wave information and the temperature distribution information. In this case, even if the shape of the air-conditioned space is changed due to the rearrangement of furniture or the renovation of the air-conditioned space, learning data can be obtained based on the updated room shape.

In step S43, the obtainer 30 or the target area determiner 35 determines whether the determination time has elapsed from the time point at which the obtainer 30 obtains radio wave information for the first time. If the determination time has elapsed (YES in step S43), the air-conditioning controller 3 completes the obtaining processing for radio wave information. If the determination time has not elapsed (NO in step S43), the air-conditioning controller 3 returns to step S41. In step S42, which is re-executed after step S43, the obtainer 30 cumulatively stores information, such as radio wave information and temperature distribution information, without updating those stored in the storage 31 by newly obtained information, such as radio wave information and temperature distribution information. Learning data may include the position information based on radio wave information at one time point, in which case, step S43 may be omitted.

FIG. 14 is a flowchart illustrating an example of learning processing executed by the learning model generator according to Embodiment 2. In step S51, the learning model generator 350 obtains radio wave information, temperature distribution information, and room shape information from the storage 31. The room shape information stored in the storage 31 may be information obtained together with radio wave information or information obtained in advance.

In step S52, the learning model generator 350 obtains position information based on the radio wave information. In step S53, the learning model generator 350 generates a learning model based on learning data indicating the position information, temperature distribution information, and room shape information. In step S54, the learning model generator 350 stores the generated learning model in the storage 31. After step S54, the learning model generator 350 completes the learning processing.

FIG. 15 is a flowchart illustrating an example of target area determination processing executed by the air-conditioning controller according to Embodiment 2. In step S61, the obtainer 30 obtains temperature distribution information. In S61, the obtainer 30 may also obtain room shape information. In step S62, the estimator 351 estimates a user stay area from the temperature distribution information obtained in step S61 based on the learning model and determines a target area. In step S62, the estimator 351 may determine a target area from the room shape information as well as the temperature distribution information. The room shape information may be room shape information obtained in step S61 or room shape information obtained by the obtainer 30 and stored in the storage 31 in advance. After step S62, the air-conditioning controller 3 completes the target area determination processing.

In Embodiment 2, the obtainer 30 obtains radio wave information indicating the radio wave intensity and the radio wave receiving direction. However, if one or both of the remote controller 12 and the terminal device of a user has a global positioning system (GPS) function (hereinafter such a device may also be called a GPS-equipped device), for example, the obtainer 30 may obtain, from this GPS-equipped device, position information indicating this device. Alternatively, an application program for identifying the position of the GPS-equipped device in the air-conditioned space may be installed in this device, and the obtainer 30 may obtain position information from this device. In these cases, learning data indicating certain information including position information is stored in the storage 31, and the learning model generator 350 generates a learning model from the learning data stored in the storage 31 without obtaining position information based on radio wave information. In this case, radio wave information in steps S41 through S43 is replaced by position information. Radio wave information in step S51 is also replaced by position information, and step S52 is omitted.

In Embodiment 2, the learning model generator 350 generates a learning model from temperature distribution information obtained from the single sensor 20 for detecting the temperature distribution in the air-conditioned space. However, the number of sensors 20 for detecting the temperature distribution in the air-conditioned space may be variable. The learning model generator 350 may generate a learning model based on temperature distribution information obtained from each of the sensors 20 for detecting the temperature distribution in the air-conditioned space. These sensors 20 are installed at different positions in the air-conditioned space. This enables the learning model generator 350 to obtain more detailed information on the temperature distribution in the air-conditioned space. The learning model generator 350 can thus generate a more precise learning model, thereby improving the accuracy in estimation processing for the position of a user executed by the estimator 351.

In Embodiment 2, the learning model generator 350 generates a learning model from room shape information obtained from the single sensor 20 for detecting the shape of the room in the air-conditioned space. However, the number of sensors 20 for detecting the shape of the room in the air-conditioned space may be variable. The learning model generator 350 may generate a learning model based on room shape information obtained from each of the sensors 20 for detecting the shape of the room in the air-conditioned space. These sensors 20 are installed at different positions in the air-conditioned space. This enables the learning model generator 350 to obtain more detailed information on the shape of the room in the air-conditioned space. The learning model generator 350 can thus generate a more precise learning model, thereby improving the accuracy in estimation processing for the position of a user executed by the estimator 351.

In Embodiment 2, the learning model generator 350 generates a learning model from radio wave information or position information obtained from the single remote controller 12. However, the number of remote controllers 12 may be variable. The learning model generator 350 may generate a learning model based on radio wave information or position information obtained from each of the remote controllers 12. With this configuration, if there are multiple users in the air-conditioned space, the learning model generator 350 can relate the positions of the individual users obtained from signals output from the corresponding remote controllers 12 to the positions of the corresponding heat sources indicated by the temperature distribution information. The learning model generator 350 can thus generate a learning model from such learning data, from which the estimator 351 can estimate the position of each user with higher accuracy.

In Embodiment 2, the learning model generator 350 generates a learning model from radio wave information or position information obtained from the single terminal device. However, the number of terminal devices may be variable. The learning model generator 350 may generate a learning model based on radio wave information or position information obtained from each of the terminal devices. With this configuration, if there are multiple users in the air-conditioned space, the learning model generator 350 can relate the positions of the individual users obtained from signals output from the corresponding terminal devices to the positions of the corresponding heat sources indicated by the temperature distribution information. The learning model generator 350 can thus generate a learning model from such learning data, from which the estimator 351 can estimate the position of each user with higher accuracy.

In Embodiment 2, the air-conditioning system 100 includes one air-conditioning controller 3. However, the number of air-conditioning controllers 3 in the air-conditioning system 100 may be variable. The air-conditioning system 100 may include multiple air-conditioning controllers 3. In this case, the learning model generator 350 of one of the air-conditioning controllers 3 may each generate a learning model based on certain information, such as temperature distribution information, obtained by all of the air-conditioning controllers 3. Then, the learning model generated by the learning model generator 350 of this air-conditioning controller 3 may be stored in the other air-conditioning controllers 3. In this case, the other air-conditioning controllers 3 may each estimate the position of a user based on this learning model or may update the learning model by conducting relearning. Alternatively, an external learning model generator 350 disposed outside the multiple air-conditioning controllers 3 may each generate a learning model based on certain information, such as temperature distribution information, obtained by all of the air-conditioning controllers 3. The learning model generated by the external learning model generator 350 may be stored in the multiple air-conditioning controllers 3. In this case, each air-conditioning controller 3 may estimate the position of a user based on this learning model or may update the learning model by conducting relearning.

In Embodiment 2, by way of example, the learning model generator 350 generates a learning model by using a neural network formed by input layers, intermediate layers, and output layers. However, the learning model generator 350 may generate a learning model based on deep learning. Alternatively, the learning model generator 350 may conduct machine learning by genetic programing, functional logic programing, or a support vector machine, for example.

Advantages achieved by the air-conditioning controller 3 according to Embodiment 2 will be described below. The occupant stay information in Embodiment 2 is temperature distribution information indicating the temperature distribution in the air-conditioned space. The obtainer 30 receives a radio wave from the remote controller 12 or the terminal device, which is used for operating the air-conditioning apparatus 1, and obtains radio wave information indicating the intensity and the receiving direction of the radio wave. The target area determiner 35 includes the estimator 351. The estimator 351 determines the target area based on a learning model, which is obtained from the radio wave information and the temperature distribution information and is used for estimating the area where an occupant is present in the air-conditioned space. The position of the remote controller 12 or the terminal device that has emitted the radio wave can be associated with the position of a user. Hence, with the use of the radio wave information, an occupant can be identified from among heat sources indicated by the temperature distribution information. The air-conditioning controller 3 can thus highly accurately determine the area including the position of an occupant from the occupant stay information as the target area, based on the learning model obtained from the radio wave information and the temperature distribution information.

The target area determiner 35 in Embodiment 2 further includes the learning model generator 350. The learning model generator 350 obtains position information indicating the position of the remote controller 12 or the terminal device based on the radio wave information, and generates a learning model that is used for estimating the area where an occupant is present in the air-conditioned space, based on the position information and the temperature distribution information. Because of this learning model, the estimator 351 can estimate the position of the occupant with high accuracy from the occupant stay information. The target area determiner 35 can thus highly accurately determine the target area, in which the occupant is present. This also improves the accuracy of the objective function determined by the integration of deviations in the target area. Therefore, the air-conditioning controller 3 can execute the air conditioning optimization efficiently.

The occupant stay information in Embodiment 2 is temperature distribution information indicating the temperature distribution in the air-conditioned space. The obtainer 30 obtains, from a remote controller 12 or a terminal device, which is used for operating the air-conditioning apparatus 1, position information indicating the position of the remote controller 12 or the terminal device. The target area determiner 35 includes the estimator 351. The estimator 351 determines the target area based on a learning model, which is obtained from the position information and the temperature distribution information and is used for estimating the area where an occupant is present in the air-conditioned space. The position of the remote controller 12 or the terminal device that has emitted the radio wave can be associated with the position of a user. Hence, with the use of the position of the remote controller 12 or the terminal device, an occupant can be identified from among heat sources indicated by the temperature distribution information. The air-conditioning controller 3 can thus highly accurately determine the area including the position of an occupant from the occupant stay information as the target area, based on the learning model obtained from the radio wave information and the temperature distribution information.

The target area determiner 35 in Embodiment 2 further includes the learning model generator 350. The learning model generator 350 generates a learning model that is used for estimating the area where an occupant is present in the air-conditioned space, based on the position information and the temperature distribution information. Because of this learning model, the estimator 351 can estimate the position of the occupant with high accuracy from the occupant stay information. The target area determiner 35 can thus highly accurately determine the target area, in which the occupant is present. This also improves the accuracy of the objective function determined by the integration of deviations in the target area. Therefore, the air-conditioning controller 3 can execute the air conditioning optimization efficiently.

Embodiment 3

The air-conditioning controller 3 according to Embodiment 3 will be described below in detail. The elements of Embodiment 3 similar to those in Embodiment 1 and Embodiment 2 will be designated by the same reference signs. An explanation of the properties such as configurations and functions in Embodiment 3 similar to those in Embodiment 1 and Embodiment 2 will be omitted unless there are special circumstances.

The configuration of the air-conditioning system 100 according to Embodiment 3 is illustrated in FIG. 1 by way of example, as in Embodiment 1 and Embodiment 2. The configuration of the air-conditioning controller 3 according to Embodiment 3 is illustrated in FIG. 2 by way of example, as in Embodiment 1 and Embodiment 2. The hardware configuration of the air-conditioning controller 3 according to Embodiment 3 is illustrated in FIG. 9 by way of example, as in Embodiment 1 and Embodiment 2.

In Embodiment 1, the coupled analysis unit 36 derives a state variable vector at each time point within the first time range and derives an adjoint variable vector at each time point from the state variable vector at the corresponding time point. At this time, since the coupled analysis unit 36 derives the adjoint variable vector by conducting inverse analysis, it is required to store the state variable vector at each time point in the storage 31 or the above-described first memory 42 of the coupled analysis unit 36 and to read the state variable vector. In this case, however, the data amount in the first memory 42 is increased, which may delay processing, or the space of the first memory 42 may become short. To solve this problem, the coupled analysis unit 36 in Embodiment 3 executes the following processing.

FIG. 16 is a diagram for explaining processing for deriving an adjoint variable and sensitivity executed by the coupled analysis unit according to Embodiment 3. The thick white arrow extending straight in FIG. 16 indicates the direction of the flow of processing executed by the coupled analysis unit 36. The coupled analysis unit 36 first conducts the above-described CFD simulation to derive the state variable vector within the first time range along the forward direction of time. Then, the coupled analysis unit 36 selects multiple time points within the first time range and stores the state variable vectors at the selected time points in the first memory 42. A predetermined time interval is provided between the selected time points. Hereinafter, each of the time points may also be called a check time point.

Among the state variable vectors at the individual time points t within the first time range obtained by the above-described CFD simulation, the coupled analysis unit 36 stores the state variable vector at the check time point t_m in the first memory 42. It is defined that m is an integer of 0 to n and n is an integer of 1 or greater. In FIG. 16, within the first time range from the analysis start time point t₀ to the end time point (t₀+T), (n+1) check time points t_m are selected, and the state variable vector at each check time point t_m is stored in the first memory 42.

Along the backward direction of the check time point t_m, the coupled analysis unit 36 conducts forward analysis and inverse analysis to derive the sensitivity at each time point t in the period between the check time point t_m and the check time point t_m+1. This will be explained below in detail.

As the first forward analysis, the coupled analysis unit 36 derives the state variable vector at each time point in the period from the check time point t_n-1 to the check time point t_n. At this time, the coupled analysis unit 36 conducts CFD simulation based on the already obtained indoor model including the state quantity at each of the check time points t_n-1 and t_n. As the first inverse analysis, the coupled analysis unit 36 derives the adjoint variable vectors at the individual time points t in the period from the check time point t_n to the check time point t_n-1, based on the state variable vectors at the corresponding time points t in the period from the check time point t_n-1 to the check time point t_n. As the first sensitivity deriving processing, the coupled analysis unit 36 derives the sensitivity at the individual check time points t_n-1 to t_n, based on the state variable vectors at the corresponding time points t in the period from the check time point t_n-1 to the check time point t_n and the adjoint variable vectors at the corresponding time points t in the period from the check time point t_n to the check time point t_n-1.

As the second forward analysis, the coupled analysis unit 36 derives the state variable vector at each time point t in the period from the check time point t_n-2 to the check time point t_n-1. At this time, the coupled analysis unit 36 conducts CFD simulation based on the already obtained indoor model including the state quantity at each of the check time points t_n-2 and t_n-1. As the second inverse analysis, the coupled analysis unit 36 derives the adjoint variable vectors at the individual time points t in the period from the check time point t_n-1 to the check time point t_n-2, based on the state variable vectors at the corresponding time points t in the period from the check time point t_n-2 to the check time point t_n-1. As the second sensitivity deriving processing, the coupled analysis unit 36 derives the sensitivity at the individual check time points t_n-2 to t_n-1, based on the state variable vectors at the corresponding time points t in the period from the check time point t_n-2 to the check time point t_n-1 and the adjoint variable vectors at the corresponding time points t in the period from the check time point t_n-1 to the check time point t_n-2.

As the final n-th forward analysis, the coupled analysis unit 36 derives the state variable vector at each time point in the period from the check time point t₀ to the check time point t₁. As the n-th inverse analysis, the coupled analysis unit 36 derives the adjoint variable vector at each time point in the period from the check time point t₁ to the check time point t₀. As the n-th sensitivity deriving processing, the coupled analysis unit 36 derives the sensitivity at the individual check time points t₀ to t₁, based on the state variable vectors at the corresponding time points in the period from the check time point t₀ to the check time point t₁ and the adjoint variable vectors at the corresponding time points in the period from the check time point t₁ to the check time point t₀.

FIG. 17 is a flowchart illustrating an example of sensitivity deriving processing executed by the coupled analysis unit in Embodiment 3. As processing in step S13 in FIG. 5, the coupled analysis unit 36 executes processing in FIG. 17 instead of processing in FIG. 6. Processing in FIG. 17 may be executed in parallel with step S12 or after step S11 and before step S12. For easy understanding, in FIG. 17, as well as in FIG. 16, the check time point is indicated by t_m, the number of check time points is (n+1), and the check time point t_m is t₀ to t_n.

In step S71, the coupled analysis unit 36 selects (n+1) check time points t₀ to t_n within the first time range. In step S72, the coupled analysis unit 36 stores the state quantity at each check time point among the state quantities derived in step S11 described above in the first memory 42. In step S73, the coupled analysis unit 36 inputs n−1 to m. In step S74, the coupled analysis unit 36 conducts CFD simulation to derive the state quantity at each time point in the period from the check time point t_m to the check time point t_m+1.

In step S75, the coupled analysis unit 36 conducts inverse analysis to derive the adjoint variable at each time point in the period from the check time point t_m+1 to the check time point t_m, based on the corresponding state quantity obtained in step S74. In step S76, the coupled analysis unit 36 derives the sensitivity at each time point in the period from the check time point t_m to the check time point t_m+1 from the corresponding state quantity obtained in step S74 and the corresponding adjoint variable obtained in step S75.

In step S77, the coupled analysis unit 36 determines whether m is 0. If m is not 0 (NO in step S77), the coupled analysis unit 36 replaces m by (m−1) in step S78. That is, the coupled analysis unit 36 reduces the number of m by one. After step S78, the coupled analysis unit 36 proceeds to step S74. If m is 0 (YES in step S77), the coupled analysis unit 36 completes the sensitivity deriving processing.

Advantages achieved by the air-conditioning controller 3 according to Embodiment 3 will be described below. The coupled analysis unit 36 in Embodiment 3 selects multiple check time points within the first time range. Based on the state quantities at the selected check time points obtained by forward analysis, the coupled analysis unit 36 conducts forward analysis and inverse analysis between the check time points along the backward direction of the check time points. By conducting forward analysis between the check time points, the coupled analysis unit 36 derives the state quantity at each time point between the check time points. Then, by conducting inverse analysis between the check time points, the coupled analysis unit 36 derives the adjoint variable of the state quantity derived at each time point between the check time points. The coupled analysis unit 36 then derives the sensitivity at the individual time points between the check time points, based on the state quantities and the adjoint variables at the corresponding time points between the check time points. The sensitivity is a quantity representing a degree of influence on the objective function when the value of the control parameter is changed. The coupled analysis unit 36 derives the optimal solution based on the sensitivity. With this configuration, instead of all of the state quantities in the first time range, the coupled analysis unit 36 stores the state quantities at the check time points in the first memory 42 and, based on the stored state quantities at the corresponding check time points, the coupled analysis unit 36 can derive the sensitivity at the individual time points between the check time points. As a result of deriving the sensitivity at each time point between the check time points along the backward direction of the check time points, the coupled analysis unit 36 can obtain the sensitivity at each time point of the first time range. It is thus possible to secure the storage space of the first memory 42 required for deriving the sensitivity. Hence, even a device having small computational resources can carry out calculation required for deriving desired control.

Embodiment 4

The air-conditioning controller 3 according to Embodiment 4 will be described below in detail. The elements of Embodiment 4 similar to those in Embodiment 1 through Embodiment 3 will be designated by the same reference signs. An explanation of the properties such as configurations and functions in Embodiment 4 similar to those in Embodiment 1 through Embodiment 3 will be omitted unless there are special circumstances.

FIG. 18 is a block diagram illustrating an example of the configuration of an air-conditioning system according to Embodiment 4. The air-conditioning system 100 according to Embodiment 4 includes a server 6 in addition to the air-conditioning apparatus 1, the detector 2, and the air-conditioning controller 3. The air-conditioning controller 3 of Embodiment 4 corresponds to the air-conditioning controller 3 of each of Embodiment 1 through Embodiment 3 from which the coupled analysis unit 36 is removed and to which a first communication unit 39 is added. The first communication unit 39 performs wired communication or wireless communication with the server 6.

The server 6 includes a second communication unit 60, which performs wired communication or wireless communication with the air-conditioning controller 3. In place of the air-conditioning controller 3, the server 6 in Embodiment 4 includes the above-described coupled analysis unit 36, which executes processing operations in FIGS. 5, 6, and 17 illustrated by way of example.

The first communication unit 39 sends the indoor model constructed by the indoor model constructor 33, the device model constructed by the device model constructor 34, and information indicating the target area determined by the target area determiner 35 to the server 6. The second communication unit 60 of the server 6 receives the indoor model, device model, and information indicating the target area. As in Embodiment 1 through Embodiment 3, the coupled analysis unit 36 of the server 6 executes processing in FIG. 5 and processing in FIG. 6 or 17, based on the indoor model, device model, and information indicating the target area received by the second communication unit 60, thereby deriving the optimal solution of the control variable vector. The second communication unit 60 sends the optimal solution derived by the coupled analysis unit 36 to the air-conditioning controller 3. The first communication unit 39 of the air-conditioning controller 3 receives the optimal solution based on the server 6.

The control target determiner 37 of the air-conditioning controller 3 derives the control target value in a manner similar to Embodiment 1, based on the optimal solution received by the first communication unit 39 from the server 6. More specifically, the control target determiner 37 determines, as the control target value, the temperature at a position such as the first position on the temperature distribution in the air-conditioned space achieved by the optimal solution. The commander 38 controls the air-conditioning apparatus 1 based on the control target value determined by the control target determiner 37.

FIG. 19 is a diagram illustrating an example of the hardware configuration of the air-conditioning controller according to Embodiment 4. The air-conditioning controller 3 according to Embodiment 4 includes a first communication interface circuit 46 in addition to the elements represented by the hardware configuration shown in FIG. 9. The functions of the first communication unit 39 can be implemented by the first communication interface circuit 46.

FIG. 20 is a diagram illustrating an example of the hardware configuration of the server according to Embodiment 4. The server 6 can be formed by, for example, a second processor 62, a second memory 63, and a second communication interface circuit 64 connected to each other via a second bus 61. The second processor 62 is a CPU or an MPU, for example. The second memory 63 is a ROM or a RAM, for example.

The functions of the coupled analysis unit 36 in Embodiment 4 can be implemented as a result of the second processor 62 reading and executing various programs stored in the second memory 63. The functions of the second communication unit 60 can be implemented by the second communication interface circuit 64. All or some of the functions of the server 6 may be implemented by dedicated hardware.

Advantages achieved by the air-conditioning system 100 and the air-conditioning controller 3 according to Embodiment 4 will be described below. The air-conditioning system 100 according to Embodiment 4 includes the air-conditioning controller 3 and the server 6. The air-conditioning controller 3 controls the air-conditioning apparatus 1, which performs air conditioning for an air-conditioned space. The server 6 communicates with the air-conditioning controller 3. The air-conditioning controller 3 includes the obtainer 30, the indoor model constructor 33, the first communication unit 39, the control target determiner 37, and the commander 38. The obtainer 30 obtains room shape information indicating the shape of the air-conditioned space and air state information indicating the state of air in the air-conditioned space. The indoor model constructor 33 constructs an indoor model used for CFD simulation, based on the room shape information and the air state information. The first communication unit 39 sends the indoor model to the server 6. The server 6 includes the coupled analysis unit 36 and the second communication unit 60. The coupled analysis unit 36 executes CFD simulation based on the indoor model to derive a state quantity indicating the state of air at each time point within a predetermined first time range. Then, the coupled analysis unit 36 derives a deviation of the state quantity at each time point from a preset target state quantity and integrates the deviations at the individual time points within the first time range. The coupled analysis unit 36 then derives an optimal solution of a control parameter of the air-conditioning apparatus 1 such that an objective function including the integrated value obtained by integrating the deviations is minimized. The second communication unit 60 sends the optimal solution to the air-conditioning controller 3. The control target determiner 37 of the air-conditioning controller 3 determines a control target value, which is a target value of the control parameter, based on the optimal solution obtained from the server 6. The commander 38 controls the air-conditioning apparatus 1 based on the control target value.

With the above-described configuration, analysis processing for deriving the optimal solution is executed in the server 6. Thus, even if the computational performance of the air-conditioning controller 3 is limited, the optimal solution can still be obtained, thereby enabling appropriate air-conditioning control. Additionally, the optimal solution can be obtained promptly according to the computational performance of the server 6, thereby making it possible to speedily improve the analyticity of the user.

While the embodiments have been described above, the present disclosure is not limited to the above-described embodiments and encompasses a conceivable range of equivalents. The configurations discussed in Embodiment 1 through Embodiment 4 and modified examples of the embodiments can be combined with each other as long as the functions and operations of the present disclosure are not impaired.

REFERENCE SIGNS LIST

• • 1: air-conditioning apparatus, 2: detector, 3: air-conditioning controller, 5: telecommunication line, 6: server, 10: outdoor unit, 11: indoor unit, 12: remote controller, 20: sensor, 30: obtainer, 31: storage, 32: current state estimator, 33: indoor model constructor, 34: device model constructor, 35: target area determiner, 36: coupled analysis unit, 37: control target determiner, 38: commander, 39: first communication unit, 40: first bus, 41: first processor, 42: first memory, 43: storage device, 44: input interface circuit, 45: input-output interface circuit, 46: first communication interface circuit, 60: second communication unit, 61: second bus, 62: second processor, 63: second memory, 64: second communication interface circuit, 100: air-conditioning system, 350: learning model generator, 351: estimator, BA: user stay area, CA: candidate area, H: user, Q_RAC: air-conditioning capacity, X1, X2, X3: input layer, Y1, Y2: intermediate layer, Z1, Z2, Z3: output layer, w11 to w16, w21 to w26: weight, Ω: target area