BOREHOLE DATA FILTERING TECHNIQUES

Description

BACKGROUND

This section is intended to introduce the reader to various aspects of art that may be related to various aspects of the present disclosure, which are described and/or claimed below. This discussion is believed to be helpful in providing the reader with background information to facilitate a better understanding of the various aspects of the present disclosure. Accordingly, it may be understood that these statements are to be read in this light, and not as admissions of prior art.

Sometimes, boreholes will cut through natural fracture in the shape of planes. As image logs are viewed as unfolded cylinders, the natural fractures get the form of a sinusoid with a period of the width of the borehole image. These sinusoids may include any amplitude, phase, and vertical position in the well. The period is, however, for physical reasons, always the width of the image. In the case of natural fractures, these sinusoids show up as dark lines on brighter backgrounds. There are, however, other sinusoids. A boundary between two formations might show up as sinusoids separating two areas of different darkness, and resistive fractures could show up as bright sinusoids on a darker background in electromagnetic (EM) images. It may be desirable to develop techniques to detect sinusoids and other features that may result in unexpected operation of boreholes.

BRIEF DESCRIPTION

Certain embodiments commensurate in scope with the originally claimed invention are summarized below. These embodiments are not intended to limit the scope of the claimed invention, but rather these embodiments are intended only to provide a brief summary of possible forms of the invention. Indeed, the invention may encompass a variety of forms that may be similar to or different from the embodiments set forth below.

In certain embodiments, a method includes receiving, via a processing system, borehole data comprising a sinusoid-like pattern. The method also includes receiving, via the processing system, one or more reference features from a storage component. Further, the method includes filtering, via the processing system, the borehole data based on the sinusoid-like pattern and the one or more reference features to generate a first filtered borehole data and a first set of false detections. Even further, the method includes determining, via the processing system, one or more additional features corresponding to false detections based on the filtered borehole data and the first set of false detection. Even further, the method includes filtering, via the processing system, the first filtered borehole data based on the one or more additional features to generate a second filtered borehole data and a second set of false detections.

In certain embodiments, a system includes a computing system comprises one or more processors. The system also includes a memory storing instructions that, when executed by the computing system, are configured to cause the computing system to perform operations including receiving borehole data comprising a sinusoid-like pattern. The instructions, when executed by the computing system, also cause the processor to perform operations including receiving one or more reference features from a storage component. Further, the instructions, when executed by the computing system, cause the processor to perform operations including filtering the borehole data based on the sinusoid-like pattern and the one or more reference features to generate a first filtered borehole data and a first set of false detections. Further still, the instructions, when executed by the computing system, cause the processor to perform operations including determining one or more additional features corresponding to false detections based on the filtered borehole data and the first set of false detections. Further still, the instructions, when executed by the computing system, cause the processor to perform operations including filtering the first filtered borehole data based on the one or more additional features to generate a second filtered borehole data and a second set of false detections.

In certain embodiments, one or more tangible non-transitory computer-readable memory media include processor-executable instructions that, when executed by one or more processors, cause the one or more processors to receive borehole data including a sinusoid-like pattern; receive one or more reference features from a storage component; filter the borehole data based on the one or more reference features and the sinusoid-like pattern to generate a first filtered borehole data and a first set of false detections; determine one or more additional features corresponding to false detections based on the filtered borehole data and the first set of false detections; and filter the first filtered borehole data based on the one or more additional features to generate a second filtered borehole data and a second set of false detections.

BRIEF DESCRIPTION OF THE DRAWINGS

These and other features, aspects, and advantages of the present disclosure will become better understood when the following detailed description is read with reference to the accompanying drawings in which like characters represent like parts throughout the drawings, wherein:

FIG. 1 illustrates a drilling system for drilling an earth formation, in accordance with the present disclosure;

FIG. 2 illustrates a block diagram of a borehole image analysis system that may be used to analyze borehole images captured for the drilling system of FIG. 1 as described in greater detail herein, in accordance with the present disclosure;

FIG. 3 is a block diagram of an example process for filtering borehole data, in accordance with the present disclosure;

FIG. 4A shows a first image of borehole data that is a false detection, in accordance with the present disclosure;

FIG. 4B shows a second image of borehole data that is a false detection, in accordance with the present disclosure;

FIG. 5A shows an image with a sinusoid-like pattern, in accordance with the present disclosure;

FIG. 5B shows the image of FIG. 5A after flattening, in accordance with the present disclosure;

FIG. 6A shows an image of a borehole data that includes a line artifact, in accordance with the present disclosure;

FIG. 6B shows a first image of identified regions within the borehole data of FIG. 6A, in accordance with the present disclosure;

FIG. 6C shows a second image of identified regions within the borehole data of FIG. 6A, in accordance with the present disclosure;

FIG. 7A shows an example of a left-handed spiral pattern, in accordance with the present disclosure;

FIG. 7B shows an example of a right-handed spiral pattern, in accordance with the present disclosure; and

FIG. 8 shows borehole data that includes a spiral pattern and corresponding artifact scores, in accordance with the present disclosure.

DETAILED DESCRIPTION

One or more specific embodiments of the present disclosure will be described below. In an effort to provide a concise description of these embodiments, all features of an actual implementation may not be described in the specification. It should be appreciated that in the development of any such actual implementation, as in any engineering or design project, numerous implementation-specific decisions must be made to achieve the developers' specific goals, such as compliance with system-related and business-related constraints, which may vary from one implementation to another. Moreover, it should be appreciated that such a development effort might be complex and time consuming, but would nevertheless be a routine undertaking of design, fabrication, and manufacture for those of ordinary skill having the benefit of this disclosure.

When introducing elements of various embodiments of the present invention, the articles “a,” “an,” “the,” and “said” are intended to mean that there are one or more of the elements. The terms “comprising,” “including,” and “having” are used in an open-ended fashion, and thus should be interpreted to mean “including, but not limited to.” Also, any use of any form of the terms “connect,” “engage,” “couple,” “attach,” or any other term describing an interaction between elements is intended to mean either an indirect or a direct interaction between the elements described. In addition, as used herein, the terms “axial” and “axially” generally mean along or parallel to a central axis (e.g., central axis of a body or a port), while the terms “radial” and “radially” generally mean perpendicular to the central axis. For instance, an axial distance refers to a distance measured along or parallel to the central axis, and a radial distance means a distance measured perpendicular to the central axis. The use of “top,” “bottom,” “above,” “below,” and variations of these terms is made for convenience but does not require any particular orientation of the components.

As mentioned above, ensuring borehole stability is desirable for maintaining expected operation and the efficiency of oil well drilling. One indicator of borehole stability are fractures, such as sinusoidal fractures. Certain conventional techniques for identifying sinusoidal fractures determine a presence of sinusoidal fractures based on a detected sinusoid in borehole data. However, it is presently recognized that, in some instances, a sinusoid may not be a fracture. Further, certain fractures may have a shape that deviates from an ideal sinusoid, and as such, may not be detected. Moreover, it is presently recognized that the quality of the borehole data (e.g., borehole image or image), the presence of artifacts (e.g., borehole data artifacts), and other features may reduce the success of determining actual sinusoid fractures. As such, determining a sinusoidal fracture is present because a sinusoid is present in the borehole data may lead to numerous false detections (e.g., false negatives and/or false positives. Accordingly, it is presently recognized that it is advantageous to develop techniques for filtering our false detections from borehole data.

Accordingly, this disclosure relates to techniques for filtering out false detections (e.g., false positives and/or false negatives) by determining features and/or artifacts in borehole data that are indicative of false detections. In general, the techniques include receiving borehole data (e.g., candidate solutions, an initial set of borehole data), and filtering out a first set of false detections from the borehole data to generate a first filtered borehole data (e.g., corresponding to the borehole data that are potential correct detections). As described herein, the borehole data may generally include a sinusoid-like pattern (e.g., a set of pixels within borehole data or an image) that potentially corresponds to a sinusoid fracture. For example, the sinusoid-like pattern may extend along an axis (e.g., an x-axis or y-axis) of borehole data, have a peak or amplitude, or otherwise appear to have a wave-like shape. In some embodiments, the borehole data may be an output of a sinusoid detection model. In any case, the sinusoid-like pattern may not correspond to a sinusoid fracture, but instead it may be an artifact, variation in intensity, or otherwise improperly labeled. In some instances, to filter out the first set of false detections, the techniques include determining whether one or more features and/or artifacts are present in the initial set of borehole data. For example, a processor may receive, retrieve, or otherwise obtain one or more reference features of detected sinusoids that are correlated with false detections. As described in further detail herein, the one or more features may include intensity related features, image quality related features, data quality features, noise and structure related features, artifacts, or a combination thereof. The first set of false detections and the first filtered borehole data may be used as reference borehole data to identify one or more additional features (e.g., extracted features) that may be used by the processor to determine whether the feature in the borehole data corresponds to a sinusoidal fracture. Then, the filtered set of borehole data may be further filtered using the one or more additional features to generate an additional filtered set of borehole data and a second set of false detections. In some embodiments, the extracted features may be provided as input to a model, using various approaches as described herein such as machine learning model techniques, neural network techniques, and the like. In this way, false detections may be removed from systems or models that utilize borehole data to determine control adjustments based on detected sinusoids. As such, computational resources (e.g., computer memory) may be utilized for other operations as opposed to being used for analyzing borehole data that corresponds to false detections.

With the foregoing in mind, FIG. 1 shows one example of a drilling system 100 for drilling a geological formation 101 to form a borehole 102. The drilling system 100 includes a drill rig 103 used to support and rotate a drilling tool assembly 104 that extends downward into the borehole 102. The drilling tool assembly 104 may include a drill string 105, a bottomhole assembly (“BHA”) 106, and a bit 110, attached to the downhole end of drill string 105.

The drill string 105 may include several joints of drill pipe 108 connected end-to-end through tool joints 109. The drill string 105 transmits drilling fluid through the borehole 102 and transmits rotational power from the drill rig 103 to the BHA 106. In some embodiments, the drill string 105 further includes additional components, such as subs, pup joints, and so forth. The drill pipe 108 provides a hydraulic passage through which drilling fluid is pumped from the surface. The drilling fluid discharges through nozzles, jets, or other orifices in the bit 110 and/or the BHA 106 for the purposes of cooling the bit 110 and cutting structures thereon, and for transporting cuttings out of the borehole 102.

The BHA 106 may include the bit 110 or other components. An example BHA 106 may include additional or other components (e.g., coupled between to the drill string 105 and the bit 110). Examples of additional BHA components include drill collars, stabilizers, measurement-while-drilling (“MWD”) tools, logging-while-drilling (“LWD”) tools, downhole motors, underreamers, section mills, hydraulic disconnects, jars, vibration or dampening tools, other components, or combinations of the foregoing downhole well tools. The bit 110 may also include other cutting structures in addition to or other than a drill bit, such as milling or underreaming tools. In general, the drilling system 100 may include other drilling components and accessories, such as make-up/break-out devices (e.g., iron roughnecks or power tongs), valves (e.g., kelly cocks, blowout preventers, and safety valves), other components, or combinations of the foregoing. Additional components included in the drilling system 100 may be considered a part of the drilling tool assembly 104, the drill string 105, or a part of the BHA 106 depending on their locations in the drilling system 100.

The bit 110 in the BHA 106 may be any type of bit suitable for degrading formation or other downhole materials. For instance, the bit 110 may be a drill bit suitable for drilling the geological formation 101. Example types of drill bits used for drilling earth formations are fixed-cutter or drag bits, roller cone bits, and percussion hammer bits. In some embodiments, the bit 110 is an expandable underreamer used to expand a wellbore diameter. In other embodiments, the bit 110 is a mill used for removing metal, composite, elastomer, other downhole materials, or combinations thereof. For instance, the bit 110 may be used with a whipstock to mill into a casing 107 lining the borehole 102. The bit 110 may also be used to mill away tools, plugs, cement, and other materials within the borehole 102, or combinations thereof. Swarf or other cuttings formed by use of a mill may be lifted to surface, or may be allowed to fall downhole.

FIG. 2 is a block diagram of a borehole image analysis system 250 that may be used to analyze borehole images captured for the drilling system 100 of FIG. 1 as described in greater detail herein. The borehole images may be received from a camera 202 as input data 252 at a computing system 254. In certain embodiments, the camera 202 may be implemented within a downhole tool (e.g., as part of a BHA 106) that may transmit the input data 252 as wired or wireless communications. In certain embodiments, the downhole tool that includes the camera 202 may include any components discussed in relation to the computing system 254. Indeed, in certain embodiments, the computing system 254 may be and/or may include the downhole tool that includes the camera 202. However, in other embodiments, the computing system 254 may be located at the surface of the drilling system 100 of FIG. 1. The various functional blocks shown in FIG. 2 may include hardware elements (including circuitry), software elements (including computer code stored on a tangible computer-readable medium), or a combination of both hardware and software elements. It should be noted that FIG. 2 is merely one example of a particular implementation and is intended to illustrate the types of components that may be present in the computing system 254.

As illustrated, the computing system 254 may include one or more processor(s) 256, a memory 258, a display 260, input devices 262, one or more neural networks(s) 264, and one or more interface(s) 266. In the computing system 254, the processor(s) 256 may be operably coupled with the memory 258 to facilitate the use of the processors(s) 256 to implement various stored programs. Such programs or instructions executed by the processor(s) 256 may be stored in any suitable article of manufacture that includes one or more tangible, computer-readable media at least collectively storing the instructions or routines, such as the memory 258. The memory 258 may include any suitable articles of manufacture for storing data and executable instructions, such as random-access memory, read-only memory, rewritable flash memory, hard drives, and optical discs. In addition, programs (e.g., an operating system) encoded on such a computer program product may also include instructions that may be executed by the processor(s) 256 to enable the computing system 254 to provide various functionalities. For example, the processor 256 may utilize reference features to determine whether a sinusoid-like pattern in borehole data corresponds to a sinusoid fracture or whether the sinusoid-like pattern is an artifact or other non-sinusoid pattern. At least in some instances, the processor 256 may utilize the relationships between features and false detections to train, generate, or refine a model used to detect sinusoid fractures.

The input devices 262 of the computing system 254 may enable a user to interact with the computing system 254 (e.g., pressing a button to increase or decrease a volume level). The interface(s) 266 may enable the computing system 254 to interface with various other electronic devices. The interface(s) 266 may include, for example, one or more network interfaces for a personal area network (PAN), such as a Bluetooth network, for a local area network (LAN) or wireless local area network (WLAN), such as an IEEE 802.11x Wi-Fi network or an IEEE 802.15.4 wireless network, and/or for a wide area network (WAN), such as a cellular network. The interface(s) 266 may additionally or alternatively include one or more interfaces for, for example, broadband fixed wireless access networks (WiMAX), mobile broadband Wireless networks (mobile WiMAX), and so forth.

In certain embodiments, to enable the computing system 254 to communicate over the aforementioned wireless networks (e.g., Wi-Fi, WiMAX, mobile WiMAX, 4G, LTE, and so forth), the computing system 254 may include a transceiver (Tx/Rx) 267. The transceiver 267 may include any circuitry that may be useful in both wirelessly receiving and wirelessly transmitting signals (e.g., data signals). The transceiver 267 may include a transmitter, a receiver, or a transmitter and a receiver combined into a single unit.

The input devices 262, in combination with the display 260, may allow a user to control the computing system 254. For example, the input devices 262 may be used to control/initiate operation of the neural network(s) 264 and/or machine learning models. Some input devices 262 may include a keyboard and/or mouse, a microphone that may obtain a user's voice for various voice-related features, and/or a speaker that may enable audio playback. The input devices 262 may also include a headphone input that may provide a connection to external speakers and/or headphones.

In certain embodiments, the neural network(s) 264 may include hardware and/or software logic that may be arranged in one or more network layers. In certain embodiments, the neural network(s) 264 may be used to implement machine learning and may include one or more suitable neural network types. For instance, the neural network(s) 264 may include a perceptron, a feed-forward neural network, a multi-layer perceptron, a convolutional neural network, a long short-term memory (LSTM) network, a sequence-to-sequence model, and/or a modular neural network. In some embodiments, the neural network(s) 264 may include at least one deep learning neural network.

As discussed below, the output of the neural network(s) 264 may be based on the input data 252, such as metrics used to identify sinusoids in borehole images, as described in greater detail herein. This output may be used by the computing system 254. Additionally or alternatively, the output from the neural network(s) 264 and/or the processor(s) 256 may be transmitted using a communication path 268 from the computing system 254 to a gateway 270. The communication path 268 may use any of the communication techniques previously discussed as available via the interface(s) 266. For instance, the interface(s) 266 may connect to the gateway 270 using wired (e.g., Ethernet) and/or wireless (e.g., IEEE 802.11) connections. The gateway 270 couples the computing system 254 to a wide-area network (WAN) connection 272, such as the Internet. The WAN connection 272 may couple the computing system 254 to a cloud network 274. The cloud network 274 may include one or more computing systems 254 grouped into one or more locations (e.g., data centers). The cloud network 274 includes one or more databases 276 that may be used to store the output of the neural network(s) 264. Indeed, in some embodiments, the camera 202 (or the computing device that the camera 202 is a part of) may send the input data 252 to the cloud 274 via connection 278 (e.g., Wi-Fi, cellular, and/or Internet connections). In such embodiments, the computing system 254 may be implemented in the cloud 274. Additionally or alternatively, at least some of the processing may be performed in the computing device that includes the camera 202. In some embodiments, the cloud network 274 may perform additional transformations on the data using its own processor(s) 256 and/or neural network(s) 264. As such, all of the following steps discussed as performed in the computing system 254 may be performed in a computing system 254 that includes the camera 202 (e.g., computing device 206), a computing system 254 that is separate from but receives image/video from the camera 202, the cloud network 274, and/or any other suitable computing devices.

As discussed herein, it may be advantageous to determine features (e.g., reference features and/or extraction features) that may be used by a processor to determine whether borehole data including sinusoid-like patterns are false detections. For example, the determined features may be used to filter borehole data used for sinusoid fracture detection such that computational resources are not utilized on borehole data having a sinusoid-like pattern that is a false detection. To generally illustrate this process, FIG. 3 shows a data flow diagram of an example computer implemented method 300 for removing false detections 302 from borehole data 304. In general, certain process blocks performed in the method 300 may be performed by the processor 256 of the computing system 254. Moreover, certain process blocks described below may be performed in a different order than that illustrated, and, indeed, in some embodiments, certain process blocks may be skipped altogether. For simplicity, the method 300 is described as being performed by the computing system 254.

As shown, the computing system 254 receives borehole data 304. In general, the borehole data 304 includes a detected sinusoid-like pattern. In some embodiments, the computing system 254 may receive the borehole data from the camera 202 as input data 252. In some embodiments, the computing system 254 may receive the borehole data 304 from a downhole tool (e.g., part of the BHA 106 described in FIG. 1). In some embodiments, the computing system 254 may receive the borehole data from a storage component, such as the one or more databases 276, the memory 258, or another suitable storage component.

At block 306, the computing system 254 filters the borehole data 304 to generate false detections 302 (e.g., a first set of false detections 302). In general, the false detections may be false positives, false negatives or both. In general, the computing system 254 may filter borehole data 304 using reference data features 308. As shown, the reference data features 308 are stored within the database 276. However, the reference data features 308 may be stored in any suitable storage component, such as the memory 258 of the computing system 254, a cloud database, and so on.

The reference data features 308 may facilitate the computing system 254 in detecting or identifying artifacts and/or features in the borehole data 304 that correspond to false detections. In some embodiments, the reference features may include patterns (e.g., data patterns, pixel patterns) or mathematical relationships for identifying features, such as intensity related features, image quality related features, noise and structure related features, or a combination thereof. In some embodiments, the reference features may include patterns for detecting artifacts, such as a missing value artifact, a line artifact, a salt and pepper artifact, a stretched data artifact, a spiral artifact, or a combination thereof. These features and artifacts are discussed in more detail with respect to block 314.

As one non-limiting example, the reference features 308 may include a mean intensity of the image along the trace of the predicted sinusoid: For example, it is presently recognized that natural fractures may include a solid band of relatively low pixel intensity (e.g., less than or equal to 50%, 40%, 30%, 20%, or 10% of the peak intensity). As such, this feature may differentiate a sinusoid corresponding to a natural fracture from other types of sinusoidal patterns. For example, a sinusoid corresponding to a natural fracture may have a dark band that contains a distribution of different intensities, darker than their surrounding areas.

As another non-limiting example, the reference features 308 may include a true dip feature, such as a true dip threshold: As referred to herein, the “true dip” is the angle of inclination of a layer (e.g., geographic layer) with respect to the horizontal plane in the geographic frame. Mechanically speaking, the fractures appear orthogonal to the rock layers. As such, it is presently recognized that if the layers are horizontal then the fractures are vertical, and therefore the natural fractures have a relatively high dip true value (e.g. greater than or equal to 50, 55, 60, 65, 70, or 75). In such an embodiment, the computing system 254 may determine a dip true value of the sinusoid-like feature. If the true dip value is below a threshold, then the computing system 254 may determine that the sinusoid-like pattern does not correspond to a sinusoid fracture. However, if the true dip value exceeds the threshold, then the computing system 254 may determine that the sinusoid-like pattern does correspond to the sinusoid fracture.

As another non-limiting example, the reference feature 308 may include a threshold number of pixels having a relatively low intensity: As mentioned above, the natural fractures have a solid band of low intensity. It is presently recognized that if a sinusoid-like pattern includes a number of connected pixels under a certain threshold of intensity (e.g., intensity threshold), the computing system 254 may determine that the sinusoid-like pattern is an artifact rather than a sinusoid corresponding to a sinusoid fracture.

At block 312, the computing system 254 generates references images. In general, it is presently recognized that it may be advantageous to, given the original images and their corresponding ground truth, create small patches or subsets from all detections of the initial algorithm, and divide them into true positives (the sinusoids that correspond to sinusoid fractures) and false positives (the patterns that were detected, but do not correspond to sinusoid fractures). It is presently recognized that true positives and false positives may be useful for determining features that can be significant and discriminant to differentiate the real fractures from everything else. Additionally, they may assist in training a machine learning model designed to differentiation between the two scenarios.

The patches themselves were made to show all sinusoidal patterns into one standard format, regardless of their initial phase or amplitude. In general, patches are portions of the whole image (e.g., borehole image or borehole data). The computing system 254 may use the borehole data containing sinusoids to construct a dataset with true positives (i.e., patches containing a sinusoid). Further, the computing system 254 may use the borehole data that does not contain sinusoids to generate an additional dataset with false positives (i.e., patches not containing sinusoids but misclassified by the algorithm as sinusoids). This may be done by the computing system 254 phase shifting each pattern to the same position, leaving a predetermined amount of background above and below for better context. Below is an example of patches containing fractures before and after transformation for both cases true and false positives. Note how the scale of y-axis differ, although the sinusoids are displayed in the same format.

At block 312, the computing system 254 generates extracted features 316 (e.g., additional features). In general, the extracted features 316 may be used to further filter false detections from the borehole data 304. At least in some instances, the extracted features 316 may be stored in the database 276, thereby updating the reference features 308 such that subsequent filtering (e.g., as described in block 306) may utilize new features (e.g., the extracted features 316). In this way, the disclosed techniques may reduce the likelihood of false detections from remaining in the borehole data 304 and being used to sinusoid fracture detection.

As described herein, the reference features 308 may include patterns or mathematical relationships for identifying features that are correlated with false detections, such as intensity related features, image quality related features, noise and structure related features, artifacts, or a combination thereof. Further, the extracted features 316 may also include patterns or mathematical relationships for identifying features that are correlated with false detections, such as intensity related features, image quality related features, noise and structure related features, or a combination thereof.

Intensity Related Features

As one non-limiting example, the reference features 308 and/or extracted features 316 may include a minimum index and minimum intensity value. As described herein, a sinusoid-like pattern (e.g., a sinusoid pattern, a sinusoid fracture pattern) that corresponds to a sinusoid fracture may exhibit relatively low intensity. As such, it is presently recognized that flattening the sinusoid-like pattern (e.g., using the processor) corresponding to the sinusoid fracture may align the sinusoid-like pattern centrally within the borehole data 304, appearing as a straight line or nearly so. Thus, when a sinusoid-like pattern (e.g., a subset of pixels within borehole data 304) includes an actual sinusoid, the minimum intensity index in the row axis may approximate the midpoint (e.g., midpoint value) of the image. As one example of a process for flattening, the computing system 254 may establish or identify a margin around the midpoint of the sinusoid-like pattern (e.g., potential sinusoid fracture pattern) as at least one feature for analysis and filtering. Additionally, given the generally low intensity of fractures, it may be advantageous that the computing system 254 implement a threshold (e.g., intensity threshold, pixel intensity threshold) on the mean intensity around this central line. The computing system 254 may compute the mean over the columns (e.g., pixels along the y-axis) then the minimum is supposed to be in the middle of the flattened curve if the feature corresponds to a sinusoid fracture. Within wishing to be bound by theory, the minimum intensity value may occur at this midpoint. Otherwise, the computing system 254 may determine the feature does not correspond to a sinusoid fracture.

As one non-limiting example, the reference features 308 and/or extracted features 316 may include the mean intensity of the image. For example, the computing system 254 may determine the minimum intensity value along the sinusoid trace. For patches containing fractures, the mean intensity around the feature is expected to be significantly lower compared to patches without fractures. Accordingly, if the mean intensity around the feature is below a threshold (e.g., less than about 90%, 80%, 70%, 60%, 50%, 40%, 30%, or 20% of a subset of pixels not corresponding to the feature.)

As one non-limiting example, the reference features 308 and/or extracted features 316 may include the contrast of the image. For example, given the distinctive dark color of the fracture or sinusoid-like pattern, it is presently recognized that there may be a notable disparity between the fracture and the background. Without wishing to be bound by theory, the contrast plays a fundamental role in visual perception and can serve as a crucial factor in filtering out false predictions. Accordingly, the computing system 254 may determine an amount of contrast between one or more pixels of the feature and one or more pixels outside of the feature.

Image Quality Related Features

As one non-limiting example, the reference features 308 and/or extracted features 316 may include a number of missing values. For example, it is presently recognized that missing value artifacts may result in a processor falsely determining that a sinusoid-like pattern corresponds to a sinusoid fracture, when the sinusoid-like pattern is actually a set of pixels having missing values. As referred to herein, a “missing value artifact” in data refers to the absence of values for certain areas in image log. The missing value artifact may occur for various reasons, such as data entry errors, equipment malfunctions, and so on. It is recognized that missing values may be identified or noticed by the computing system 254. For example, the missing value artifact might stand out as unusual, out of distribution, null values such as “nan” or “−9999” within the image

Detecting missing value artifacts that blend into the background is relatively more difficult due to, for example, the similarity in the intensity of missing values to background in certain portions of the borehole data. However, it is presently recognized that missing value artifacts may exhibit shapes (e.g., rectangles, lines, etc.) that are not expected to be present in borehole data.

One specific non-limiting example of detecting missing value artifacts is described below. For example, applying the one or more reference features to the borehole data when the one or more reference features include missing value patterns (e.g., to detect missing value artifacts) includes applying a thresholding technique to the borehole data. For example, the thresholding technique may convert pixels having a value above or below a threshold to a single number. For example, if the range of values of borehole data are between 1-100, all pixel values 70 or greater, 80 or greater, 90 or greater, 95 or greater, may be converted to 100. This may emphasize bright areas, which may facilitate detection of missing value artifacts. Moreover, the computing system 254 may identify clusters of pixels based on the similarity of the pixel values. As such, the clusters may be the regions discussed herein.

Additionally or alternatively, applying the one or more reference features to the borehole data when the one or more reference features include missing value patterns includes filtering out isolated pixels and regions of the resulting image that may not correspond to missing value artifacts. For example, the computing system 254 may determine a region does not correspond to a missing value artifact if the shape of the region does not match the missing value pattern above or below a threshold. Several techniques could be employed for this purpose, including contour detection, morphological operations, and line detection. It is presently recognized that morphological operations offer high flexibility in detecting different shapes, based on the choice of the kernel. As referred to here, the “kernel” refers to a modeling component (e.g., software modeling component) that may be defined according to the form of the object (e.g., the artifact) that is desirable to detect. For example, to detect an artifact having a rectangular shape, it may be advantageous to define the kernel as a rectangle. For example, the new kernel may be selected to capture dark lines more efficiently and filter out all unnecessary detected areas. In some embodiments, the kernel is a horizontal line, but with different shape proportions to detect only very large horizontal objects. By selecting the appropriate kernel type (e.g., square, line, rectangle), the computing system 254 may target specific regions of pixels in the borehole data having a shape that corresponds to missing value artifacts.

After removing regions that do not correspond to the missing value artifacts, the computing system 254 may determine an artifact score for each region. For example, the computing system 254 may determine the artifact score for a missing value artifact (e.g., a missing value artifact score) based on the percentage of total missing values per row.

As one non-limiting example, the reference features 308 and/or extracted features 316 may include artifacts, such as line artifacts. As such, the computing system 254 may determine whether the borehole data 304 includes an artifact by comparing a reference pattern (e.g., a line artifact pattern) to the borehole data 304 or otherwise identifying a collection of pixels that may include missing values. If the reference data pattern matches the feature or the missing values otherwise occur within the feature, the computing system 254 may determine that the feature is actually an artifact. As referred to herein, a “line artifact” in data refers to non-geological features present in borehole images. In some embodiments, to detect a line artifact, the computing system 254 may apply a thresholding technique to the borehole data. For example, the thresholding technique may convert pixels having a value above a threshold to a single number. In some instances, it may be advantageous to preserve all or relatively most of the relevant dark information present in the image.

To determine whether the sinusoid-like feature is a line artifact, the computing system 254 may determine a score indicative of the match between the sinusoid-like feature and a reference feature 308 (e.g., a reference line artifact). As such, the computing system 254 may determine that the borehole image 304 having the feature is a false detection 302 when the score exceeds a threshold score.

Further, after applying the thresholding technique, the computing system 254 may apply a filtering technique to preserve horizontal lines. However, the kernel may be different. For example, the new kernel may be selected to capture dark lines more efficiently and filter out all unnecessary detected areas. In some embodiments, the kernel is a horizontal line, but with different shape proportions to detect only very large horizontal objects. After detecting potential line artifacts, the computing system 254 may apply an error rate criterion to determine whether to confirm the presence of line artifacts in the labeled borehole data 304. For example, the values of this artifact score are either 0 (no artifacts) or 1 (artifacts).

Accordingly, in some embodiments, to detect a missing value artifact and/or a line artifact, the computing system 254 may apply a thresholding technique to borehole data, thereby producing or generating modified borehole data (e.g., adjusted borehole data). Further, the computing system 254 may identify one or more regions within the modified borehole data. Then, the processor may determine whether a shape of the one or more regions matches a reference data 308, such as a reference missing value artifact and/or missing value artifact data pattern having a shape that corresponds to a particular artifact.

As one non-limiting example, the reference features 308 and/or extracted features 316 may include spiraling patterns or artifacts. For example, borehole spiraling is a common artifact in borehole images, which show up as diagonal lines over the image and can lead to errors in interpretation. To detect the spiral artifacts, the computing system 254 may identify regions of the borehole data having characteristics of a sinusoid depth, phase, and amplitude. Non-limiting examples of techniques to identify a spiral (e.g., using a spiral template or kernel) may include performing two-dimension (2D) convolutions with cylindrical padding, pattern recognition and/or identification using 2D convolutions, sinusoid feature (e.g., depth, phase, amplitude, or a combination thereof) matching, self-interaction pattern searching, or a combination thereof.

In some embodiments, the computing system 254 may utilize a spiral kernel or a sinusoid kernel to identify an artifact. In some embodiments, to detect the spiral artifact, the computing system 254 may apply a right-handed pattern and a left-handed pattern corresponding to a spiral pattern to the borehole data. Further, the computing system 254 may detect the one or more regions based on at least a partial match between the right-handed pattern or the left-handed pattern. In some embodiments, there may be no limitation if the actual spirals are shifted along the x axis or has a steeper or shallower slope than the patterns. As such, the computing system 254 may rescale the pattern along the y-axis as well as shift it along the x-axis, to detect any such fit. Certain techniques may be unable to distinguish between left-handed and right-handed slopes. As such, it may be advantageous to use both the right-handed pattern and the left-handed pattern corresponding to a spiral pattern.

To detect the spiral pattern, in some embodiments, the computing system 254 may run the patterns over the borehole image, which may lead to two similarity matrices s, telling and s_l, with “r” and “l” for left-handed and right-handed. These matrices are 3-dimensional and may indicate the fit of the pattern for each depth, phase, and amplitude.

In an embodiment the phase or of the amplitude are not used, the computing system 254 may only utilize the maximum similarity score of each depth, updating s_r and s_l so that they become 1-dimensional with one value/true vertical depth (tvd). Although described in terms of tvd, it should be noted that the processor 256 may utilize a maximum similarity score of each depth, updating s_r and s_l so that they become 1-dimensional with one value/measured depth (md) or another type of depth measurement. It is presently recognized that a spiral pattern may not occur twice at the same time, as that would not be a spiral but, for example, a sinusoid that is going down on one side and coming back up on the other one. To eliminate such sections from the spiral pattern, it may be advantageous to utilize a single similarity vector s=|s_r−s_l|.

To provide a spiral score that reflects spiraling that is pervasive over some given depth, it may be advantageous to average the scores so that the score for each depth is the average over some window, for example 100 pixels. The computing system 254 may analyze the artifact scores over wells to find a threshold s_max, above which actual spiraling behavior is near certain. Then, the computing system 254 may divide the spiral scores by this value and put all values above it to 1, so that:

s ← min ⁢ ( s s max , 1 )

As such, the computing system 254 may obtain a spiral score s that provides a flag for spiraling whenever reaching 1 and indicates that there is no spiraling when close to 0. Values in-between may be seen as some intrinsic uncertainty level. As such, it may be desirable to generate an alert indicating that additional review or input is desirable.

Noise and Structure Related Features

As one non-limiting example, the reference features 308 and/or extracted features 316 may include a structural similarity index measure (SSIM). In general, the SSIM is a metric used to quantify the similarity between two images. The computing system 254 may utilize the SSIM to assess the perceived quality of an image by comparing its structural information, such as textures, edges, and other high-level details, rather than just comparing pixel values. For example, the SSIM may be utilized by the computing system 254 to consider luminance, contrast, and structure, providing a more comprehensive evaluation of image similarity. Higher SSIM values indicate greater similarity between the images, with 1 representing perfect similarity.

One non-limiting example of a technique for determining SSIM is described below. In general, the technique includes using similarity assessment to differentiate patches containing sinusoidal patterns (true positives) from those without (false positives). For example, the computing system 254 may utilize a predetermined threshold to emphasize the sinusoid pattern located in the center of the patch. The computing system 254 may determine the threshold value as the median intensity of pixels extracted from the predicted sinusoid pattern within the image. The parameters of the predicted sinusoid may be defined by a sinusoid model. As such, these techniques may identify and rectify any missed detections by other models. The resulting binary image may contain additional details, yet the sinusoidal shape may be more distinguishable in patches with true sinusoids compared to those without.

As one non-limiting example, the reference features 308 and/or extracted features 316 may include an entropy calculation. In general, the computing system 254 may utilize entropy for image analysis is to quantify the level of information and randomness present in images. It is presently recognized that images with high entropy may exhibit more information and textures, while those with lower entropy often appear smoother and more uniform. For example, borehole data 304 containing fractures tend to be more uniform, whereas some borehole data 304 with erroneous predictions display a greater amount of texture and randomness.

Connected Components

As one non-limiting example, the reference features 308 and/or extracted features 316 may include a connected component value. In general, a connected component in an image denotes a cluster of adjacent pixels that share a common property or attribute. Put differently, the connected component represents a set of pixels connected through neighboring relationships, such as 4-connectivity/Von Neumann neighborhood (e.g., horizontally or vertically adjacent pixels) or 8-connectivity/Moore neighborhood (e.g., including diagonals).

At block 318, the computing system 254 filters the first subset of borehole data 310. As such, the computing system 254 may generate a second subset of borehole data 320 and false detections 302b (e.g., corresponding to false detections that were not filtered out from the borehole data 304 at block 306). In general, the second subset of borehole data 320 may have fewer false detections as compared to the first subset of borehole data 310 and the borehole data 304. The computing system 254 may utilize the extracted features 316 to filter the borehole data 310. In this way, the disclosed techniques may provide an effective post-processing technique to eliminate as many false positives as possible. Before resorting to machine learning models for filtering, it may be advantageous to utilize a rule-based method leveraging certain extracted features to discard obvious cases, leaving more challenging ones for machine learning to handle.

In some embodiments, the rule-based method may use a ranked list of extracted features 316 and/or reference features 308. For example, the computing system 254 may rank extracted features 316 and/or reference features 308 based on an increasing likelihood that a particular feature indicates that borehole data 304 is a false detection. For example, the computing system 254 may utilize 1, 2, 3, 4, or more than 4 extracted features 316 that have the highest likelihood of detecting a false detection. As one non-limiting example, the rule-based method may include filtering using the minimum value of intensity (e.g., expected to be centered around the middle of the flattened image), the minimum value of intensity, and a number of missing values. Accordingly, if the computing system 254 determines that the sinusoid-like pattern in a borehole data 304 has a minimum value of intensity at the center of the flattened image, the computing system 254 may then utilize the minimum value intensity feature. Similarly, if the computing system 254 determines that the pixels of the sinusoid-like pattern have a pixel value below a threshold, the computing system 254 may then utilize the number of missing value feature.

This filtering may help in discarding certain instances of incorrect predictions without compromising on the accurate ones. However, ensuring that true positives are not filtered is a difficult process. Hence, another step involving machine learning methods may facilitate learning more intricate data representations.

At block 322, the computing system 254 generates, trains, and/or refines a model, such as a sinusoid fracture detection model using the second subset of borehole data 320. In some embodiments, the computing system 254 may provide one or more of the previously mentioned features as inputs to a machine learning model, treating the problem as a classification task; given a set of features, classify whether an example represents a fracture or not. At least in some instances, the borehole data 304 may be labeled, and thus, generating the model is a supervised classification problem. Numerous machine learning models are suitable for this task, such as Random Forest Classifier. However, it should be noted that any suitable machine-learning (ML) modelling techniques may be used. In some embodiments, the computing system 254 may normalize the data, bringing all features to the same scale, which may mitigate any bias stemming from differing data scales.

As described herein, a first subset of the borehole data 304 may include sinusoids corresponding to sinusoid fractures, while a second subset of the borehole data 304 may include features that may cause a model to incorrectly identify a sinusoid as corresponding to a sinusoid fracture. To further illustrate this, FIG. 4A and FIG. 4B show borehole data 304a and borehole data 304b that include a sinusoid-like pattern 334, 336. Further, each borehole data 304a and 304b is annotated with an example sinusoid trace 338 that corresponds to sinusoid fractures. As such, although a processor may identify the sinusoid-like patterns 334, 336 of FIGS. 4A and 4B as a sinusoid. Reference features may be used to determine that the sinusoid-like patterns do not correspond to sinusoid fractures. As referred to herein, a “sinusoid-like pattern” is a subset of borehole data 304 that includes a pattern having a wave-like shape. For example, sinusoid-like pattern 334 of the borehole data 304a does not include a correctly estimated amplitude (e.g., a feature). Further, while the sinusoid-like pattern 336 of the borehole data 304b is a sinusoid (e.g., approximately fits the sinusoid trace 338), the sinusoid-like pattern 336 of the borehole data 304b is not a sinusoid type that corresponds to sinusoid fractures. Accordingly, FIGS. 4A and 4B illustrate examples of features, amplitude of sinusoid, phase shift of sinusoid, relative location of the amplitudes, and the like, that may be used to identify false detections. At least in some instances, the computing system 254 may pre-process borehole data to center the sinusoid-like pattern 336 and determine whether the sinusoid-like pattern 336 corresponds to a sinusoid fracture based on the amplitude.

As described herein, the computing system 254 may identify one or more additional features, such as intensity related features. FIG. 5A shows borehole data 304 that includes a sinusoid-like pattern 340. FIG. 5B shows an adjusted borehole data 341 that includes a flattened pattern 342 (e.g., indicated by the box). In some embodiments, a fracture feature (e.g., a sinusoid fracture feature) generally includes a relatively lower intensity than other features exhibited in borehole data 304. It is presently recognized that flattening borehole data 304 (e.g., to produce the adjusted borehole data 341) that includes a sinusoid-like pattern 340 should facilitate determining whether the sinusoid-like feature 340 is a true positive or true negative. For example, if the computing system 254 adjusts a sinusoid pattern by aligning the peaks with a node, then the resulting adjusted sinusoid pattern should be a substantially horizontal line.

In some embodiments, the reference data feature 308 and/or the extracted features 316 (e.g., additional features) may be a line artifact pattern. In some embodiments, the line artifact pattern represents a range of the borehole data where an acquisition interruption may have occurred. To illustrate this, FIGS. 6A, 6B, and 6C (e.g., FIGS. 6A-6C) show a borehole data 304, an image 360, and an image 362. The borehole data 304 of FIG. 6A includes a line artifact. The image 360 of FIG. 6B shows identified regions 354a, 354b, and 354c (e.g., collectively 354). In general, the regions 354a, 354b, and 354c may be detected in a generally similar manner as described in block 306 of FIG. 3. For example, the computing system 254 may apply the reference data pattern 308 that is a line artifact pattern to the borehole data 304 and determine areas or a subset of pixels that match the line artifact pattern within a threshold similarity, such as 50% or greater match, 60% or greater match, 70% or greater match, 80% or greater match, or 90% or greater match. The image 362 of FIG. 6C illustrate regions 356a and 356b that include artifacts or otherwise correspond to regions 354 that have an artifact score above an artifact score threshold. For example, as shown, the image 352 includes region 356a that corresponds to region 354a. The image 352 does not include a region 354 that corresponds to the region 354b or region 354c. As such, region 354c and the region 354b represents regions 354 having an artifact score in a threshold range that does not correspond to the region 354 having an artifact. Accordingly, the computing system 254 may generate a label indicating that at least the region 356a (e.g., or region 354a) includes a line artifact.

As described herein, the one or more reference data features 308 may include a spiral pattern. To illustrate this, FIG. 7A shows a reference data pattern 308 that is a left-handed spiral pattern 390a. Further, FIG. 7B shows a reference data pattern 308 that is a right-handed spiral pattern 390b. In general, the computing system 254 may identify both left-handed spiral patterns and right-handed spiral patterns in a generally similar manner as described with respect to block 306 of FIG. 3. To illustrate how the left-handed spiral pattern 390a and the right-handed spiral pattern 390b may be used to identify artifacts, FIG. 8 shows borehole data 304 and a calculated spiral score log 400. In general, the calculated spiral score log 400 indicates a spiral artifact score (e.g., ‘1’ indicating the presence of the spiral artifact and ‘0’ indicating the absence of the spiral artifact). Accordingly, the computing system 254 may identify certain regions based on the spiral score being above or below a threshold value, such as 0.4, 0.5, 0.6, 0.7, 0.8, and the like.

Technical effects of this disclosure include techniques for reducing computational resources used for operations that may lead to false detections. For example, it is presently recognized that the presence of borehole data that includes sinusoid-like features that correspond to false detections may result in an inefficient use of computational resources, as a processor may improperly identify features (e.g., sinusoid fractures) using the sinusoid-like patterns that may not correspond to a sinusoid associated with a sinusoid fracture. Accordingly, by filtering borehole data and generating steps to automate identification of features useful for filtering, the disclosed techniques may prevent computational resources from being wasted, and thus the computational resources may be used to perform other tasks, such as more quickly determining the presence of other borehole features in other borehole data.

1. A method, comprising receiving, via a processing system, borehole data comprising a sinusoid-like pattern; receiving, via the processing system, one or more reference features from a storage component; filtering, via the processing system, the borehole data based on the sinusoid-like pattern and the one or more reference features to generate a first filtered borehole data and a first set of false detections; determining, via the processing system, one or more additional features corresponding to false detections based on the filtered borehole data and the first set of false detections; and filtering, via the processing system, the first filtered borehole data based on the one or more additional features to generate a second filtered borehole data and a second set of false detections.

The method of any preceding clause, further training a sinusoid fracture detection model using the second filtered borehole data, the second set of false detections, or both.

The method of any preceding clause, wherein the first set of false detections correspond to false positives.

The method of any preceding clause, wherein the one or more reference features comprise intensity related features, image quality related features, noise and structure related features, or a combination thereof.

The method of any preceding clause comprising receiving, via the processing system, the borehole data as an output from a sinusoid detection model.

The method of any preceding clause, wherein the one or more reference features correspond to one or more borehole data artifacts, one or more data quality features, or a combination thereof.

7. The method of any preceding clause, wherein filtering the borehole data comprises: flattening the borehole data along a sinusoid pattern; determining a subset of the flattened borehole data that includes a minimum intensity value at the sinusoid pattern; and outputting the subset of the flattened borehole data as the first set of borehole data.

The method of any preceding clause, further comprising training a sinusoid fracture detection model using the one or more additional features.

The method of any preceding clause, wherein filtering the borehole data comprises: determining a minimum intensity value of the sinusoid-like pattern; determining the minimum intensity value is below an intensity threshold; and outputting a subset of the borehole data as the first set of borehole data based on the minimum intensity value being below the intensity threshold.

A system, comprising a computing system comprises one or more processors and a memory storing instructions that, when executed by the computing system, are configured to cause the computing system to perform operations comprising: receiving borehole data comprising a sinusoid-like pattern; receiving one or more reference features from a storage component; filtering the borehole data based on the sinusoid-like pattern and the one or more reference features to generate a first filtered borehole data and a first set of false detections; determining one or more additional features corresponding to false detections based on the filtered borehole data and the first set of false detections; and filtering the first filtered borehole data based on the one or more additional features to generate a second filtered borehole data and a second set of false detections.

The system of any preceding clause, wherein the instructions that, when executed by the computing system, are configured to cause the computing system to train a sinusoid fracture detection model using the second filtered borehole data, the second set of false detections, or both.

The system of any preceding clause, wherein the instructions that, when executed by the computing system, are configured to cause the computing system to filter the borehole data by determining a minimum intensity value of the sinusoid-like pattern; determining the minimum intensity value is below an intensity threshold; and outputting a subset of the borehole data as the first set of borehole data based on the minimum intensity value being below the intensity threshold.

The system of any preceding clause, wherein the one or more additional features comprise missing value patterns, line artifact patterns, spiraling patterns, or a combination thereof.

The system of any preceding clause, wherein the instructions that, when executed by the computing system, are configured to cause the computing system to filter the borehole data by: flattening the sinusoid-like pattern in the borehole data; determining a plurality of midpoint values based on the flattened borehole data; determining whether the plurality of midpoint values is below a pixel intensity threshold; and outputting a subset of the borehole data as the first set of borehole data based on the midpoint values being below the pixel intensity threshold.

The system of any preceding clause, wherein the instructions that, when executed by the computing system, are configured to cause the computing system to update the one or more reference features to include the one or more extracted features.

One or more tangible non-transitory computer-readable memory media, comprising processor-executable instructions that, when executed by one or more processors, cause the one or more processors to: receive borehole data including a sinusoid-like pattern; receive one or more reference features from a storage component; filter the borehole data based on the one or more reference features and the sinusoid-like pattern to generate a first filtered borehole data and a first set of false detections; determine one or more additional features corresponding to false detections based on the filtered borehole data and the first set of false detections; and filter the first filtered borehole data based on the one or more additional features to generate a second filtered borehole data and a second set of false detections.

The one or more tangible non-transitory computer-readable memory media of any preceding clause, wherein the instructions that, when executed by the one or more processors, are configured to cause the one or more processors to train a sinusoid fracture detection model using the second filtered borehole data, the second set of false detections, or both.

The one or more tangible non-transitory computer-readable memory media of any preceding clause, wherein the instructions that, when executed by the one or more processors, are configured to cause the one or more processors to filter the borehole data by: determining whether the sinusoid-like pattern includes a missing value artifact, wherein the one or more reference features comprise a missing value artifact pattern; determining an artifact score for the borehole data include the missing value artifact; and outputting a subset of the borehole data based on the artifact score of the subset of the borehole data.

The one or more tangible non-transitory computer-readable memory media of any preceding clause, wherein the instructions that, when executed by the one or more processors, are configured to cause the one or more processors to filter the second filtered borehole data using a machine learning (ML) model.

The one or more tangible non-transitory computer-readable memory media of any preceding clause, wherein the instructions that, when executed by the one or more processors, are configured to cause the one or more processors to filter the borehole data by determining whether sinusoid-like pattern is a sinusoid corresponding to a sinusoid fracture.

The techniques presented and claimed herein are referenced and applied to material objects and concrete examples of a practical nature that demonstrably improve the present technical field and, as such, are not abstract, intangible or purely theoretical. Further, if any claims appended to the end of this specification contain one or more elements designated as “means for (perform)ing (a function) . . . ” or “step for (perform)ing (a function) . . . ”, it is intended that such elements are to be interpreted under 35 U.S.C. 112(f). However, for any claims containing elements designated in any other manner, it is intended that such elements are not to be interpreted under 35 U.S.C. 112(f).