System for generating bit wear model predictions and related methods
A machine learning-based drill bit wear model with a drill bit wear path slope curve algorithm improves drilling efficiency by filtering noise and optimizing parameters, addressing the challenges of inconsistent predictions and equipment failures in drill bit wear.
Patent Information
- Application Number
- SA1120258382
- Authority / Receiving Office
- SA · SA
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-11-10
- Publication Date
- 2026-10-05
AI Technical Summary
Drilling operations face inefficiencies due to unpredictable drill bit wear, leading to potential equipment loss and increased costs from irreparable failures, as existing models struggle with limited and noisy field data, resulting in inconsistent predictions and suboptimal drilling parameters.
A machine learning-based drill bit wear model using a drill bit wear path slope curve algorithm to project and decode latent space representations, enhancing prediction accuracy and stability by filtering out data noise, allowing for optimized drilling parameters.
The model provides more reliable and efficient drilling operations by predicting drill bit wear accurately, reducing the risk of equipment failure and optimizing drilling parameters, thereby extending the lifespan of drill bits and preventing costly losses.
Smart Images

Figure 1BC8BF78-D952-4D90-B205-BF0E283C133B 
Figure 76838925-F28D-40F7-91E6-36720AC43089 
Figure 7DB4111C-114D-4425-A98E-28DADB3E1667
Abstract
Description
Drilling a wellbore may involve using a drill string at the end of a long drill pipe, commonly referred to as the drill string. The drill string can be used to drill through various formations, such as rock, dirt, sand, tar, etc. In some cases, the drill string may be configured to drill through additional elements present at the bottom of the well, such as cement, casing (e.g., bottom hole casing), discarded or lost equipment (e.g., picks, scrap, etc.), packing tools, etc. Each of these elements will cause wear and tear on different parts of the drill string as a result of the drilling process. As different parts of the drill string wear down, the likelihood increases that at least one part will fail, sometimes resulting in part or all of the drill string being ejected from the drill string—an irreparable loss within the wellbore. This can lead to significant time wasted and the loss of expensive equipment, potentially reducing drilling efficiency and necessitating the replacement of all costly drilling equipment. Therefore, to prevent drill bits from breaking in the wellbore, a machine learning-based drill bit wear model can be developed to predict bit wear, allowing for repair or replacement before in-well failure. Furthermore, by creating a drill bit wear model, various operating parameters of the drill string can be optimized to reduce the drill bit's wear rate during operation and double the drill bit's drilling distance before at least one part of the bit needs repair or replacement. Typically, this drill bit wear model might be generated using field data collected from real-time field trials. However, due to physical limitations, it may be impossible to measure drill bit wear beyond the first few feet of the wellbore. Consequently, for a drilling operation spanning several hundred or thousands of feet, the measured field data may be limited to the first few feet and the periphery of the entire drilling operation.For example, field data are typically limited to the acute start state of the earth drill bit (e.g., the first few feet when the earth drill bit is drilling into the ground) and the slack end state (e.g., the last few feet after the drill bit has been drilled into the ground and removed from the well). Furthermore, field data may exhibit significant noise (e.g., sharp variations in detected wear) or be of poor quality (e.g., measurement inaccuracies), which can result in an inconsistent drill bit wear model that oscillates between predicted acute and slack states over the length of the expected wellbore. This makes it more difficult to predict wear or optimize various drill string parameters to maximize drilling efficiency. According to this finding, the Earth Drilling Tool System (ETS) may use a machine learning-based drill bit wear model to predict drill bit wear by employing a drill bit wear path slope curve algorithm to improve the stability and accuracy of the drill bit wear model's predictions. For example, the ETS may calculate a drill bit wear path slope curve based on the data used to train the drill bit wear model. When using the drill bit wear model to make subsequent predictions based on new drilling data, the ETS may encode the new drilling data into a latent space, project the latent space representation of the new drilling data onto the drill bit wear path slope curve, and then decode the latent space representation of the predicted new drilling data to make a drill bit wear prediction. Because the ETS in this finding is capable of projecting the drill bit wear model's input data onto the drill bit wear path slope curve, the ETS outperforms its counterparts in conventional drill bit wear model generation systems.For example, by projecting the latent space representation of the input data onto a regression curve appropriate to the training data, a drill bit erosion model may produce more stable and reliable results by eliminating outliers that might have resulted from poor data quality or field data noise in the data used to train the drill bit erosion model, while still respecting and implementing the results of the data used to train the drill bit erosion model. Furthermore, the ability to produce this more accurate, reliable, and stable predictive model may enable more efficient drilling operations and help prevent the loss of expensive downhole equipment. In other words, the current exploration earthwork tool system outperforms its conventional earthwork tool counterparts because it allows for the training of a drill bit wear model based on measured field data and the acquisition of a drill bit wear path slope curve that is appropriate to the training data, thereby improving the reliability and stability of drill bit wear predictions. For example, the model may be based, at least in part, on a decoder-encoder model. Once the model is trained, the earthwork tool system may receive new drilling data (e.g., real-time drilling data). This new drilling data can then be mapped to a latent space defined by the trained drill bit wear model. A smooth drill bit wear slope curve can then be calculated based on the data represented in the latent space used to train the model. The new drilling data can then be projected onto the slope curve. The drilling data predicted by the drill bit wear model can then be "decoded" to generate a drill bit wear prediction.Therefore, given the increased stability and reliability of the drill bit wear model, the Earth Drilling Tool System may optimize one or more drilling parameters based on the drill bit wear model to enhance the performance of one or more drill bits included within the system. For example, the Earth Drilling Tool System may optimize several parameters for one or more drill bits to enable the drill bits to operate longer before requiring repair or replacement due to wear caused by drill bit operation, compared to conventional systems. Figure 1 illustrates a 100 Earth Drilling Tool System according to one or more current exploration application models. A 100 Earth Drilling Tool System may include a 102 drill string. It may also include electronics, such as sensors 114, a sensor module 116, and / or tool control components 118. The 102 drill string may consist of multiple sections of drill pipe joined together to form a long string of drill pipe. The front end of the 102 drill string may include a bottom hole assembly (BHA) 104. The BHA 104 may include components such as a motor 106 (e.g., a mud motor), one or more bore-expanders 108 and / or clamps 110, and an Earth Drilling Tool 112 such as a drill bit. Drill string 102 may be inserted into wellbore 120. Wellbore 120 may be formed by earth drilling tool 112 as the drill string progresses through formation 122. Tool control components 118 may be configured to control an operational aspect of earth drilling tool 112.For example, the 118 tool control components might include a routing component configured to change the directional angle of the 112 earth drill string relative to the 102 drill string and alter the direction of the 102 drill string's progress. The 118 tool control components might also be configured to receive instructions from the surface operator and execute actions based on those instructions. In some application models, downhole control instructions can be inferred from the 118 tool control components, such as a closed-loop system. Sensors 114 can be configured to collect information related to Earth Drilling Tool 112, such as tool temperature, cutter temperature, cutter wear, weight on bit (WOB), torque on bit (TOB), string rotational speed (RPM), and wear (e.g., deterioration, dulling, or damage) of at least one part of Earth Drilling Tool 112. The information collected by Sensors 114 can be processed and / or stored and / or transmitted by Sensor Module 116. For example, Sensor Module 116 may receive information from Sensors 114 in the form of raw data, such as voltage (e.g., DC voltage 0-10, DC voltage 0-5, etc.), current intensity (e.g., 0-20 mA, 4-20 mA, etc.), or current resistance (e.g., resistance temperature detector (RTD)). Thermistor, etc.). Sensor Module 116 may process the raw sensor data and send the data to the surface via a communications network, using a communications network protocol to transmit the raw sensor data.The communication network might include, for example, a communication line, a pressure pulse telemetry system in the mud, an electromagnetic telemetry system, a wired pipe, etc. In some application models, Sensor 116 might be configured to run calculations on baseline data extracted by the sensors, including, for example, calculating various wear and / or damage signs for Earth Drill 112 for a known undamaged, unworn, or initial condition of Earth Drill 112. Although the 114 sensors in Figure 1 are depicted as being mounted at or near the wellhead, they may be mounted anywhere (e.g., anywhere on the well or located directly on the 102 drill string or the 112 earth drill) so that the sensors are able to detect the wear of at least one part of the drill string (e.g., the 112 earth drill) in addition to field drilling data and drilling log data. In some application models, downhole information can be transmitted to the operator on the surface or to a surface computing device. For example, downhole information can be presented to the operator via a display screen, in a printed copy, etc. In other application models, downhole information may be transmitted to a computing device that processes the information and presents it to the operator in various formats useful to the operator. For example, measurements that are out of range may be provided as alerts, warning lights, alarms, etc. Some information may be provided directly via a display screen or in a spreadsheet, etc., while other information may not be useful until further calculations are performed, and the result of the calculation may be presented via the display screen, in a printed copy, a spreadsheet, etc. In some application models, Sensor 114 may be configured to capture image data representing different parts of the Earth Drilling Tool 112 and provide the image data to the display screen.The operator can then measure, calculate, or otherwise obtain the wear parameters of the earth drilling tool 112 based on the provided image data or other data collected by the sensor 114. In some application models, information obtained from sensor 114 can be used to generate one or more drill bit wear models that indicate the expected wear conditions of the Earth Drill 112 at a distance drilled in formation 122. In some application models, one or more drill bit wear models can be used to determine when at least one part of the Earth Drill 112 needs to be repaired or replaced. In some additional application models, one or more drill bit wear models can be used to optimize various drill string operating parameters for the drilling operation. For example, the Earth Drill 100 system might automatically adjust one or more drilling parameters, including but not limited to, one or more rates of penetration (ROP), drilling fluid flow rate, weight on drill bit (WOB), rotational speed (RPM), wellbore geometry, drilling fluid composition, etc., in response to the resulting drill bit wear model.In some additional application models, the Earth Drilling Tool 100 system may present one or more recommendations for modifying one or more drilling parameters in response to an output bit wear model. The operator may then either accept one or more of these recommendations or manually modify one or more drilling parameters based, at least in part, on the recommendations. The Earth Drilling Tool 100 system may also contain a processor and memory (discussed in more detail in Figure 7). Additionally, the Earth Drilling Tool 100 system may also have an input / output interface and a communication interface. In one or more application models, the processor contains instruction-executing hardware, such as that which comprises a computer program. Memory can be used to store data, metadata, and programs to be executed by the processor(s). The input / output interface allows the user to provide inputs to and outputs from the Earth Drilling Tool 100 system and to transmit and receive data to and from it (for example, to and from Sensor 114 via Sensor Module 116). It is worth noting that the communication interface may consist of hardware, software, or both. In any case, the communication interface may provide one or more communication interfaces (for example, packet-based communication) between the Earth Drilling Tool 100 system and one or more other computing or network devices. Figure 2 is a flowchart 200 illustrating the process of the Earth Drilling Tool System 100, which is managed by a processor to execute instructions stored on a computer-readable storage medium. For example, Figure 2 shows one or more application models of a simplified sequential flow used by the Earth Drilling Tool System 100 to train machine learning models and generate one or more predictive models (e.g., predictive algorithms) to predict one or more wear states of at least one drilling tool. As used here, the term "predictive model" may refer to a machine learning model trained to predict (e.g., estimate) the wear states of at least one drilling tool. In process 202, the Earth Drilling Tool System 100 receives initial drilling data. For example, initial drilling data may include field drilling data, log-drilling (LWD) data, and data indicating wear on at least one part of at least one drilling tool. Field drilling data may include one or more data sets corresponding to one or more wells (e.g., one or more wells for a drilling operation).Field drilling and drilling logging data may include one or more rock strengths, rate of penetration (ROP), weight on bit (WOB), revolutions per minute (RPM), wellbore geometry, formation geometry, formation density, tool geometry, formation composition, tool rotation, formation logging data (e.g., gamma ray data, density and photoelectric index data, neutron porosity, resistivity data, image data (e.g., one or more images of the borehole), acoustic data, formation pressure, formation fluid data, nuclear magnetic resonance (NMR), or seismic data while drilling (SWD). As a non-exhaustive example, the first drilling data may include labeled data representing different wear states of one or more drilling tools associated with the field drilling and drilling logging data for a particular drilling operation.In other words, the initial drilling data may represent different wear states of the drilling tool during the drilling process as a function of depth, field drilling data, and drilling log data corresponding to the drilling process. In some application models, drilling data may be acquired via one or more sensors (e.g., Sensor 114). For example, the sensor(s) may be positioned along the wellbore so that when a drilling tool (e.g., Earth Drill 112) passes through or is retrieved from the well, the sensor(s) may capture or otherwise receive data indicating wear on at least one part of the drilling tool. Alternatively, the sensor(s) may be placed on the drilling tool or other components of the Earth Drill 100 system, provided that the sensor(s) are capable of detecting wear on at least one part of a drilling tool (e.g., Earth Drill 112). One or more sensors may also capture field drilling data and log data during drilling operations. In some application models, the sensor(s) may capture image data of the drilling tool and transmit the image data (e.g., via Sensor 116) to an operator.In some application models, the sensor or sensors can be configured to measure the wear condition of the drill bit compared to a known undamaged, unworn, or initial state of the drill bit. In some application models, wear data may be acquired manually. For example, the operator might measure drill bit wear at various points during the drilling process. The operator might then provide the Earth 100 Drill Bit System with the data collected as wear data for at least one part of a drill bit. In some application models, the initial drilling data may be based, at least in part, on historical drilling data captured during previous drilling operations and stored for retrieval by the Earth 100 Drill Bit System. In operation 204, the Earth Drilling Tool System 100 may train at least one predictive bit wear model to predict one or more wear states for at least one drill bit from the Earth Drilling Tool System 100 based, at least in part, on initial drilling data. It is possible to configure at least one predictive bit wear model to encode the initial drilling data into a latent space representation of the initial drilling data. For example, the initial drilling data may include named bit wear data such that when at least one predictive bit wear model generates a predicted bit wear state based on an input, the predicted bit wear state can be compared to the actual named bit wear state corresponding to the input. At least one predictive bit wear model may then modify one or more of its properties in response to the comparison. As an unrestricted example, at least one predictive bit erosion model in the form of a decoder-encoder might accept first drilling data (e.g., field drilling data or log-drilling data) as input and correlate features of the first drilling data with a latent space representation of the first drilling data, as discussed above in Process 206. At least one predictive bit erosion model might then decode the latent space representation of the first drilling data to generate one or more predicted bit erosion states for the drilling tool based on the encoder output. The Earth Drilling Tool 100 system might then compare the predicted bit erosion states with the classified bit erosion data to generate a "proximity" score. If the proximity score is outside the acceptable range, the decoder might automatically modify one or more parameters or weights associated with one or more parameters in one or more decoder nodes and then use the first drilling data (e.g., the filtered latent space representation of the first data) to make another prediction.This process can be repeated until the proximity falls within an acceptable range. In some application models, the decompiler may be a machine learning model trained as XGboost / SVM, based on minimum distance, or using another conventional classification method. For example, the decompiler may be trained based on the classified data included in the initial drilling data. In operation 208, the Earth Drilling Tool 100 system may compare the latent area representation of the initial drilling data with the target data (e.g., target data 222) to generate one or more similarity scores. For example, target data 222 may represent field drilling data, bit erosion data, or historical drilling log data corresponding to drilling operations similar to the one yet to be performed. For example, the field drilling data, bit erosion data, or historical drilling log data may relate to the formation characteristics of the formation to be drilled. In some application models, the target data may be included within the initial drilling data (e.g., a subset of the initial drilling data).This target drilling data can be mapped to a latent space representation using at least one predictive bit erosion model for comparison with the latent space representation of at least a portion of the initial drilling data (e.g., one or more subsets of the initial drilling data) to generate one or more similarity scores. For example, the initial drilling data might include data representing different parameters (e.g., formation characteristics) of one or more historical or simulated drilling operations. The target data can be compared to the initial drilling data (e.g., one or more subsets of the initial drilling data) in each dimension of the latent space. For example, in a latent space defined by dimensions Z1 and Z2, the Earth Drilling Tool 100 system might compare the latent space distribution of the target data on Z1 with the latent space distribution of the initial drilling data on Z1 and calculate the similarity between the two distributions. This process can be repeated for dimension Z2 and for each dimension defining the latent space. This process can also be repeated for each drilling operation represented in the initial drilling data.For example, the initial drilling data might include data from a group of drilling operations. The distributions of the target data parameters (222) on each dimension of the latent space can then be compared with the distributions on each dimension of the latent space for each of the multiple drilling operations represented to calculate the degree of similarity for each successive drilling operation represented. After comparing each drilling operation's distribution with the target data distributions, the similarity score of the drilling operation represented in the latent space can be calculated. This similarity score can then be used to filter the data associated with the drilling operations represented in the latent space to retain drilling operation data that fall within a predefined threshold of proximity to the target data. Each drilling operation for the initial drilling data may have a calculated correlative similarity score that measures the similarity between the drilling operation data in question and the target data. After calculating one or more similarity scores, the Earth Drilling Tool 100 system may then filter the latent space representation of the initial drilling data to include data from one or more historical or simulated drilling operations that fall within the proximity threshold to the target data (e.g., the nearest 5, 10, 50, 100, or any other number of drilling operations represented in the latent space representation of the initial drilling data), as described in Process 210.In some application models, the Earth Drilling Tool 100 system may filter the initial drilling data in response to one or more calculated similarity scores. As an unrestricted example, the Earth Drilling Tool 100 system may filter one or more subsets of the initial drilling data and use the filtered initial drilling data to train at least one predictive bit erosion model. In this example, at least one predictive bit erosion model may map the initial drilling data to a latent space representation of the initial drilling data, but it will not use the initial drilling data to train at least one predictive bit erosion model until the initial drilling data is filtered in response to one or more similarity scores. In process 212, the Earth Drilling Tool System 100 may receive a second drilling data set via one or more sensors. Although it appears in flowchart 200 that this occurs after process 210, the current detection is not limited in this way. As an unrestricted example, the Earth Drilling Tool System 100 may receive a second drilling data set before establishing one or more similarity scores based on the target data 222. In this example, the second drilling data set may include the target data (for example, as a subset of the second drilling data set). It may also include field drilling data, log data during drilling, and data indicating wear on at least one part of the drilling tool. Furthermore, the second drilling data set may be an indicator of data captured by the Earth Drilling Tool System 100 in real time during the system's operation. For example, the second drilling data set may be received, as described above, in relation to the first drilling data set via sensors 114. In operation 214, the Earth Drilling Tool 100 system may fit a corrosion path regression curve to represent the latent space of the data used to train at least one predictive drill bit corrosion model (for example, the filtered latent space representation for the first drill data). For example, the corrosion path regression curve may define a smooth curve that fits the data (for example, field drill data and log-drill data for the first drill data) mapped to the latent space. In operation 216, the Earth Drilling Tool 100 system may apply at least one predictive corrosion model to encode the second drill data onto a latent space representation for the second drill data. In operation 218, the Earth Drilling Tool 100 system may then project the latent space representation for the second drill data onto the corrosion path regression curve. For example, the Earth Drilling Tool 100 system may project the latent space representation for the second drill data onto a point on the path regression curve closest to the latent space representation for the second drill data.In other words, the latent space representation of the second drilling data may be projected onto a point on the path slope curve that is the shortest distance in latent space between the latent space representation of the second drilling data and the path slope curve. In some applied models, the latent space representation of the second drill data may be projected onto the path slope curve in response to the latent space representation of the second drill data exceeding a predefined limit. For example, the Earth Drilling Tool 100 system may project the latent space representation of the second drill data if the latent space representation of the second drill data exceeds a predefined distance in latent space from the path slope curve (e.g., the distance in latent space between the latent space representation of the second drill data and the nearest point on the path slope curve). In this example, the Earth Drilling Tool 100 system may not project the latent space representation of the second drill data onto the path slope curve if the latent space representation of the second drill data is within the predefined distance from the path slope curve. In operation 220, the Earth Drilling Tool 100 system may generate one or more predicted wear states for at least one drill bit that respond to the predicted latent space representation of the second drilling data. For example, the Earth Drilling Tool 100 system may decode the predicted latent space representation of the second drilling data, using at least one predictive bit wear model, to generate at least one predicted wear state for at least one drill bit. In some application models, the predicted wear state for at least one drill bit may be in the form of a wear curve or gradient versus drilling depth, representing the predicted wear of at least one drill bit in the Earth Drilling Tool 100 system as a function of the drilling depth of at least one drill bit. Additionally, in one or more application models, the Earth Drilling Tool 100 system may generate a warning indicating that at least one part of the drilling tool needs to be replaced in response to at least one predicted wear condition generated in at least one drilling tool. For example, a predictive bit wear model may be configured to predict the depth at which one or more drilling tools are at increased risk of failure or excessive wear, or when the drilling tool will require replacement to mitigate these risks. As a specific, unrestricted example, the Earth Drilling Tool 100 system may detect when the drilling tool reaches a depth indicated by at least one predicted wear condition generated for at least one drilling tool, exceeding a predefined wear limit. The Earth Drilling Tool 100 system may then generate a warning stating that at least one part of the drilling tool needs to be replaced or repaired to prevent tool failure or excessive wear. In some application models, the Earth Drilling Tool system may then display some of the resulting warning to the operator via the Earth Drilling Tool system screen. In some application models, the Earth Drilling Tool 100 system may automatically modify one or more of the drilling parameters of the drill string 102 in response to the detection of an increased risk of failure or excessive wear. For example, the Earth Drilling Tool 100 system may modify one or more drilling parameters known to reduce the likelihood of damage to at least one drill string (e.g., Earth Drilling Tool 112). As a specific, but not limited, example, when the Earth Drilling Tool 100 system detects that one or more drill strings have reached a depth at which the wear condition of at least one drill bit indicates that the wear of one or more drill strings will exceed a predefined limit, the system may halt drilling operations unless this is manually overridden by an operator. For example, the warning displayed to the operator may include a selectable overriding option, whereby, if the warning is not overridden, the Earth Drilling Tool 100 system may cease operation after a predefined period of time.However, if the operator selects the override option before the predefined time period expires, the Earth Drilling Tool 100 system may continue operating without interruption. In addition to the warning, the Earth Drilling Tool 100 system may also display a selectable option to immediately stop drilling operations or modify one or more drilling operations. Furthermore, the Earth Drilling Tool 100 system may also display, alongside the warning, one or more recommendations. For example, one or more recommendations may include the need to change one or more operational drilling parameters for drill string 102, in addition to the recommended values. The operator may then, upon selecting the option to modify the operational drilling parameters for the drill string, choose to implement the recommended values or manually enter different values. Figure 3 is an illustration of the machine learning model of the encoder-decoder provided as an example (e.g., a predictive bit erosion model) 250. The encoder-decoder model may include input data 252, an encoder 254, a latent space 256, a decoder 258, and an output 260. For example, the encoder-decoder model may receive input data 252 (e.g., field drilling data, drilling log data, or bit erosion data). The input data in latent space 256 can then be encoded by encoder 254 into a compressed (e.g., reduced-dimensional) representation of input data 252. Decoder 258 can then decode the latent space representation of input data 252 to generate output 260, representing the expected wear state of a drill bit for at least one of the Earth Drilling System 100. The input and output sizes may differ. The encoder-decoder model can be trained using any traditionally known machine learning technique to produce the desired output based on a given input (e.g., supervised learning, unsupervised learning, or no-constraints).In some applied models, at least one predictive bit erosion model may be in the form of an autoencoder configured to automatically extract relevant features from a given input to map the input to a latent state representation. Although Figure 3 illustrates that the Earth Drilling Tool 100 system uses an encoder-decoder machine learning method, this system is not limited to current detection. Therefore, it may be possible to use any machine learning method as long as the Earth Drilling Tool 100 system is capable of receiving drilling data corresponding to one or more drill bits and training a predictive bit erosion model using the received drilling data to predict the erosion of one or more drill bits as a function of the distance drilled by one or more drill bits within a formation (e.g., a rock formation). Figure 4 is a graph 270 showing a set of drill bit wear condition predictions represented in the latent space. Graph 270 includes a set of drill bit wear conditions used for training (e.g., drill bit wear condition 278), which are represented in the graph as circles. Graph 270 also includes a large number of drill bit wear condition predictions (e.g., predicted drill bit wear condition 274), which are represented as triangles and are based on real-time input data. Graph 270 also includes the 272 drill bit wear path regression curve. As shown in Figure 4, it is possible to fit the 272 drill bit wear path regression curve to a number of drill bit wear conditions used for training that are represented in the latent space. After fitting the regression curve to the training data, one or more sets of second data (e.g., second drill data) may be received in real time.One or more second data sets can then be mapped to a latent space and represent a drill bit wear prediction based, at least in part, on one or more of the second input data sets. As illustrated in Figure 4, one or more real-time drill bit wear predictions represented in the latent space can be mapped onto the drill bit wear path regression curve 272 to improve model stability and accuracy when decoding to generate the predicted drill bit wear state. Figure 5 is a diagram 280 illustrating an example method for comparing latent data representations to calculate the degree of similarity, according to one or more application models. For example, the Earth Drilling Tool 100 system might calculate the similarity between the input features of the target data (e.g., estimated features of an undrilled well) and one or more input features corresponding to historical or simulated well runs (e.g., drilling operations) represented in the latent space. As shown in Figure 5, the latent space distribution of the input features' latent space representation for each dimension of the latent space can be visualized. For example, the diagram 282 of the latent space shows two dimensions, Z1 and Z2. The Earth Drilling Tool 100 system might compare the latent space distributions of different input parameters on each dimension to calculate the degree of similarity.For example, in Figure 284, the Earth Drilling Tool 100 system compares the distribution of a target dataset (e.g., data about an undrilled well plotted in the latent space) with the corresponding data from a historical drilling operation or simulation (e.g., formation characteristics) represented in the training data. The similarity score can be calculated in response to this comparison. For example, the Earth Drilling Tool 100 system might calculate the amount of space under the line representations of each dataset that is shared between the datasets across one of the latent space dimensions. The greater the shared space between the datasets, the greater the similarity reflected in the similarity score. This process can be repeated for each latent space dimension (e.g., in this case, it is repeated for Z2, as shown in Figure 286). The similarity scores for each latent space dimension can then be summed, averaged (e.g., by arithmetic mean, median, or another method of averaging), or otherwise combined to calculate the overall similarity score.This process can be repeated for each training dataset represented in the latent space to calculate the corresponding similarity score between the target dataset and each training dataset. The Earth Digging Tool 100 system can then filter the latent space to represent only the training datasets within a predefined similarity score limit (e.g., the nearest 10, 50, or 100 training datasets, without restrictions). Figure 6 is a diagram 290 illustrating several predicted drill bit wear scenarios before and after integrating the latent space projections of the input data onto a path regression curve. Diagram 290 includes a first drill bit wear graph 292 and a second drill bit wear graph 294. Both diagrams (292 and 294) illustrate several drill bit wear scenarios for at least one drilling tool as a function of depth in feet. Different levels of sharpness are illustrated in diagrams 292 and 294 through various gradients. Furthermore, the gradients are associated with numerical indices where 0-0 indicates a severe or non-wear condition expected for at least one drill bit, while 4-2 represents the highest expected wear condition for at least one drill bit. The first graph of drill bit wear 292 represents a large number of drill bit wear condition predictions derived from a predictive drill bit wear model that does not involve projecting latent space representations of the data onto a linear regression curve suitable for the training data represented in the latent space.As illustrated in Figure 6, sparse or inaccurate field data can lead to inconsistencies in the predictive capabilities of a predictive drill bit erosion model. For example, Figure 292 illustrates the instability that a predictive drill bit erosion model designed to make predictions based solely on sparse or inaccurate training data may exhibit. As shown in Figure 6, sparse or inaccurate data can lead to anomalous predictions that reflect the sparse or inaccurate training data, resulting in irregular drill bit erosion predictions. For example, irregular erosion predictions are made at depths ranging from approximately 80 feet (~25 meters) to approximately 110 feet (~34 meters). In contrast, Figure 294 represents a large number of predicted drill bit erosion scenarios involving the projection of latent space representations onto the input data in a regression curve appropriate to the training data.As illustrated in the second graph of drill bit erosion 294, by projecting the latent space representation of the input data onto a regression curve appropriate to the training data, the drill bit erosion prediction suite shows greater stability, accuracy, and consistency. It will be appreciated by those with average skill in the field that the functional elements of the examples shown here (e.g., functions, and / or operations, and / or actions, and / or processes, and / or methods) can be implemented by any suitable hardware, software, or firmware, or combinations thereof. Figure 7 shows unrestricted examples of applications of the functional elements revealed in this document. In some examples, some or all of the functional elements described here may be implemented by hardware specifically designed to implement the functional elements. Figure 7 depicts a block diagram of the 300 circuits that can be used in some examples to implement various functions, operations, procedures, and / or methods disclosed here. The 300 circuit contains one or more 302 processors (sometimes referred to in this document as "302 processors") and is operationally coupled to one or more data storage devices (sometimes referred to in this document as "306 storage"). Storage unit 306 also contains machine-executable code stored on it, as well as logic circuit 302 processors and logic circuit 304. The machine-executable code 308 contains information describing the functional elements that can be executed (e.g., performed by) the logic circuit 304. The logic circuit 304 is adapted to execute (e.g., perform) the functional elements described by the machine-executable code 308. It is necessary to consider circuit 300, when executing the functional elements described by the machine-executable code 308, as a special-purpose device configured to execute the functional elements disclosed herein.In some examples, 302 processors may perform the functional elements described by the machine-executable code 308 sequentially, or simultaneously (for example, on one or more different hardware platforms), or according to one or more parallel process streams. When executed by logic 304 circuits on processors 302, machine executable 308 code adapts the processors 302 to perform the operations described in the examples provided here. For example, machine executable 308 code might adapt the processors 302 to perform at least part of the operation or the entire operation shown in flowchart 200 in Figure 2. As another example, machine executable 308 code might adapt the processors 302 to perform some or all of the operations discussed in Figure 1. As a specific but not exhaustive example, machine executable 308 code might adapt the processors 302 to obtain a predictive drill bit wear model. As another non-restrictive example, machine executable 308 code might adapt the processors to project latent space representations of data onto a linear path regression curve. Processors 302 may include a general-purpose processor, a special-purpose processor, a central processing unit (CPU), a microcontroller, a programmable logic controller (PLC), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), another programmable logic device (PLD), a discrete gate logic or transistor, discrete hardware components, another PLD, or any combination thereof designed to perform the functions described here. A general-purpose computer containing a special-purpose computer processor is a special-purpose computer, while a general-purpose computer implements functional elements that conform to Machine 308 executable code (e.g., program code, firmware code, hardware descriptions) relevant to the current detection examples.It is worth noting that a general-purpose processor (which may also be referred to here as a host processor or simply a host) may be a microprocessor, but alternatively, 302 processors may contain any conventional processor, controller, microcontroller, or state-of-the-art machine. 302 processors may be designed with a combination of computing hardware, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in conjunction with a DSP core, or some other such configuration. In some examples, a 306-sized storage unit contains both volatile data storage (such as RAM, random-access memory) and non-volatile data storage (such as flash memory, a hard disk drive, a solid-state drive, EPROM, erasable programmable read-only memory, etc.). In some examples, 302 processors and 306-sized storage units may be integrated into a single device (such as a semiconductor device product, a system-on-a-chip (SoC), etc.). In other examples, 302 processors and 306-sized storage units may be designed in separate devices. In some examples, machine-executable code (MEC) may contain computer-readable instructions (such as program code or firmware code). For example, computer-readable instructions may be stored in MEC storage (MEC storage), accessed directly by MEC processors (MEC storage 302), and executed by MEC processors using at least MEC 304 logic circuits. Similarly, computer-readable instructions may be stored on MEC storage (MEC storage 306), transferred to a memory device (not shown) for execution, and executed by MEC processors using at least MEC 304 logic circuits. Thus, in some examples, MEC 304 logic circuits contain electrically configurable MEC 304 logic circuits. In some examples, machine-executable code (308) might describe the hardware (e.g., circuitry) that will be implemented in logic circuits (304) to perform the functional elements. This hardware can be described at various levels of abstraction, from low-level transistor layouts to high-level description languages. At a high level of abstraction, a hardware description language (HDL), such as the IEEE Standard Hardware Description Language (HDL), could be used. For example, VERILOG™, SYSTEMVERILOG™, or VLSI (VHDL™) could be used. HDL descriptions can be converted to descriptions at any of the many other abstraction levels desired. For example, a high-level description can be converted to a logic-level description such as register-transfer language (RTL), gate-level description (GL), layout-level description, or mask-level description. For instance, the precise operations to be performed by device logic circuits (e.g., gates, flip-flops, registers, etc.) for 304 logic circuits can be described in RTL and then converted by a synthesizer to a GL description. The GL description can then be converted by a mode and router to a layout-level description corresponding to the physical layout of an integrated circuit for a programmable logic device, a gate logic, a discrete transistor, discrete device components, or combinations thereof. Thus, in some examples, executable code might contain a 308 device HDL, an RTL, a GL, a mask-level description, other device descriptions, or any combination thereof. In examples where machine code 308 contains a hardware description (at any level of abstraction), a system (not shown, but including a 306 storage unit) may implement the hardware description described via the machine code 308. For example, but not limited to, 302 processors may contain a programmable logic device (e.g., FPGA or PLC), and 304 logic circuits may be electrically controlled to implement the circuitry related to the hardware description in the 304 logic circuits. Also, but not limited to, 304 logic circuits may contain wired logic fabricated by a manufacturing system (not shown, but including a 306 storage unit) according to the hardware description of the machine code 308. Regardless of whether the 308 machine-executable code contains computer-readable instructions or a hardware description, the 304 logic circuitry adapts to perform the functional elements described within the functional elements of the 308 machine-executable code when executing the functional elements of the 308 machine-executable code. It is worth noting that although the hardware description may not directly describe the functional elements, the hardware description describes the functional elements that the functional elements described by the hardware description are indirectly capable of executing. As used in this disclosure, the terms “unit” or “component” may refer to specific hardware implementations for performing unit or component actions and / or program objects or program routines that may be stored and / or executed by general-purpose hardware (e.g., computer-readable media, processing devices, etc.) of a computing system. In some examples, the various components, units, engines, and services described in this disclosure may be implemented as objects or processes executed on the computing system (e.g., as separate paths). Although some of the systems and methods described in this disclosure are generally described as executable within programs (stored on and / or executed by general-purpose hardware), specific hardware implementations or combinations of programs and specific hardware implementations are also possible and expected. According to current disclosure usage, the term "set" in relation to polyelement may contain a combination of all elements or any of the different subcompositions of some elements. For example, the phrase "a, b, c, d, or combinations thereof" may refer to any of the phrases a, b, c, or d; a combination of each of "a," b, c, and d"; and any subcomposition of the phrases "a," b, c, or d," for example, "a," b, and c; or "a, b, and d"; or "a, c, and d"; or "b, c, and d"; or "a and b"; or "a and c"; or "a and d"; or "b and c"; or "b and d"; or "c and d". The purpose of the terms used here, particularly in the attached protection elements (for example, the texts of the attached protection elements), is generally intended to be "open" terms (for example, the term "includes" should be interpreted to mean "includes, but is not limited to," the term "contains" should be interpreted to mean "contains at least," the term "includes" should be interpreted to mean "includes, but is not limited to," etc.). Furthermore, if a specific number is intended when mentioning the protected element, such an intention will be explicitly stated in the protected element itself; otherwise, no such intention exists. For example, to aid understanding, the following attached protected elements may contain the introductory phrases "at least one" and "one or more" to introduce the protected element. However, the use of such phrases should not be interpreted to mean that introducing the protected element through an indefinite article limits any particular claim containing that protected element mention to examples containing only one mention, even when the same protected element includes the introductory phrases "one or more" or "at least one" and indefinite articles (for example, the indefinite articles should be interpreted to mean "at least one" or "one or more"). The same applies to the use of definite articles to introduce the protected element. Furthermore, even if a specific number of protection elements are explicitly stated, experts in the field will understand that such a statement should be interpreted as meaning at least the stated number (e.g., the mere mention of "two mentions," without further modifications, means at least two mentions, or two or more mentions). Moreover, in those instances where a similar convention is used for "at least one of 'A, B, and C, etc." or "one or more of 'A, B, and C, etc.", the structure is generally intended to contain 'A' alone, 'B' alone, and 'C' alone, or 'A and B' together, or 'A and C' together, or 'B and C' together, or 'A, B, and C' together, etc. Furthermore, any interjection and / or any phrase representing two or more alternative terms, whether in the description, security features, or graphics, may be understood to contain one or both of the terms. For example, the phrase "A" or "B" should be understood to include the possibility of referring to "A" or "B" or "A" and "B" together. While the present disclosure here is illustrated in relation to some of the examples provided, people of ordinary skill and understanding will recognize that the present invention is not limited. Instead, many of the additions, deletions, and modifications described herein could be made in the examples without straying beyond the scope of the invention, as subsequently claimed, along with equivalent laws. Furthermore, features from one example could be combined with features from another, provided they do not exceed the scope of the invention, according to the inventor's judgment. Reference to drawings Figure 2 A - Receiving initial drilling data 202 b) Training at least one predictive drill bit erosion model 204 C - Encoding the initial drilling data for the underlying space using at least one predictive drill bit erosion model 206 D - Targeted data 222 e - Creating one or more similarity scores 208 and - representation of the latent area for filtering the initial drilling data responsive to one or more similarity scores 210 g - Receiving the second drilling data 212 H - Suitability of the erosion path slope curve to represent the latent area of the initial drilling data 214 (t) Applying at least one predictive corrosion model to encode the second drilling data on the latent space representation of the second drilling data. 216 Y - Projecting the latent area representation of the second drilling data onto the erosion path slope curve 218 K - Establish one or more expected wear conditions for at least one drilling tool that respond to the expected latent space representation of the 220 drilling data. Figure 3 A - Average B - Contrast C - Latent space D - The encrypted E - The deconstructor Figure 4 A - The potential area of drill bit erosion b - Latent dimension z2 c - Latent dimension z1 Figure 5 A - The potential area of drill bit erosion b - Latent dimension z2 c - Latent dimension z1 D - JS score: 0.18 e - JS score: 0.12 and - z1 distribution z - z2 distribution Figure 7 A - Processor(s) b) Logic circuits C - Data storage device(s) d - Machine-executable code
Claims
1. A drilling tool system, consisting of: a drill string comprising at least one drill string; one or more component sensors for sensing the wear of one or more drill strings; at least one processor; and instructions stored on at least one non-temporary, computer-readable storage medium which, when executed by at least one processor, cause the drilling tool system to: receive initial drilling data, including field drilling data, logging while drilling (LWD), and data indicating wear of at least one part of at least one drill string; train at least one predictive bit wear model to predict at least one or more drill string wear states based, at least in part, on the initial drilling data, where the predictive bit wear model is configured to encode the initial drilling data into a latent space representation of the initial drilling data; fit a wear trajectory regression curve to represent the latent space of the initial drilling data; and establish a wear state.
1. A ground drilling tool system according to protection element 1, where instructions stored on at least one computer-readable storage medium, when executed by at least one processor, cause the ground drilling tool system to: receive second drilling data, including field drilling data, logging while drilling (LWD), and data indicating wear of at least one part of at least one drilling tool via one or more sensors; use at least one predictive bit wear model to encode the second drilling data into a latent space representation of the second drilling data; and project the latent space representation of the second drilling data onto the wear trajectory regression curve.
2. A ground drilling tool system according to protection element 2, where instructions stored on the computer-readable storage medium cause the following4. A drilling tool system according to protection element 2, where instructions stored on a computer-readable storage medium, when executed by at least the processor, cause the drilling tool system to: project the latent space representation of the second drilling data onto the wear trajectory regression curve in response to the latent space representation of the second drilling data exceeding a predefined limit.
5. A drilling tool system according to protection element 4, where the predefined limit depends, at least in part, on the distance in latent space between the latent space representation of the second drilling data and the nearest point on the wear trajectory regression curve.
6. A drilling tool system according to elementProtection 2, where instructions stored on at least the computer-readable storage medium, when executed by at least the processor, cause the earth drilling tool system to: decode the predicted latent space representation of the second drilling data to generate at least one predicted wear condition for at least one drill bit.
7. Earth drilling tool system according to Protection 6, where instructions stored on at least the computer-readable storage medium, when executed by at least the processor, cause the earth drilling tool system to: modify one or more drilling parameters for at least one drill bit in response to at least one predicted wear condition.
8. Earth drilling tool system according to Protection 1, where the first drilling data includes one or more rock strengths, rate of penetration (ROP), weight on bit (WOB), rotations per minute (RPM), bit torque, formation logging data, wellbore geometry, formation geometry, and density.The configuration, engineering design, configuration structure, or rotation of the tool.
9. A drilling tool system according to Protection Element 1, where the initial drilling data includes classified data from one or more historical drilling operations.
10. A drilling tool system according to Protection Element 1, where instructions stored on at least the computer-readable storage medium, when executed by at least the processor, cause the drilling tool system to: generate one or more similarity scores that respond to a comparison of the latent space representation of the initial drilling data with the target drilling data; and filter the initial drilling data or the latent space representation of the initial drilling data that respond to one or more similarity scores.
11. A drilling tool system according to Protection Element 1, where instructions stored on at least the computer-readable storage medium, when executed by at least the processor, cause the drilling tool system to: generate a warning indicating that at least one part of the drilling tool needs to be replaced; and provide the operator, via the drilling tool system screen, with the generated warning.
12. A method for obtaining a bit wear model, consisting of: receiving initial drilling data from the Earth Drilling Tool System (ETS), including field drilling data, logging while drilling (LWD), and data indicating wear on at least one part of the drill bit; training at least one predictive bit wear model to predict at least one or more bit wear conditions based, at least in part, on the initial drilling data, where the predictive bit wear model is configured to encode the initial drilling data into a latent space representation of the initial drilling data; establishing one or more similarity scores that respond to a comparison of the latent space representation of the initial drilling data with the target drilling data; and filtering the initial drilling data or latent space representation of the initial drilling data that respond to one or more similarity scores.
13. A method according to Protection Element 12, also consisting of: receiving secondary drilling data, including field drilling data and logging while drilling dataDrilling (LWD, logging while drilling), and data indicating wear of at least one part of at least one drill bit via one or more sensors; fitting a wear trajectory regression curve to represent the filtered latent space of the first drilling data; and using at least one predictive bit wear model to encode the second drilling data into a latent space representation of the second drilling data.
14. The method according to safeguard element 13 also consists of: projecting the latent space representation of the second drilling data onto a wear trajectory regression curve; and generating one or more predicted wear states for at least one drill bit based, at least in part, on a predictive bit wear model and a wear trajectory regression curve.
15. The method according to safeguard element 14 involves generating at least one or more predicted wear states for the drill bit based, at least in part, on a predictive bit wear model.The wear trajectory regression curve establishes one or more predicted wear states for at least one drilling tool in response to the decoding of the predicted latent space representation of the second drilling data via at least one predictive bit wear model.
16. The method according to safeguard element 14, where the latent space representation of the second drilling data is projected onto the wear trajectory regression curve, involves projecting the latent space representation of the second drilling data onto the wear trajectory regression curve responsive to the latent space representation of the second drilling data that exceeds a predefined limit.
17. The method according to safeguard element 16, where the predefined limit depends, at least in part, on the distance in latent space between the latent space representation of the second drilling data and the nearest point on the wear trajectory regression curve.
18. The method according to safeguard element 12, which also consists of: receiving the third drilling data from the earth drilling tool system; and receiving the classified drilling data associated with the drilling data.Third; and validating at least one predictive bit wear model responsive to the third drilling data and the classified drilling data.
19. The method according to safeguard element 12, where the first drilling data includes classified data from one or more historical drilling operations.
20. A computer-readable, non-transitional medium storage instruction that, when executed by at least one processor, causes at least one processor to perform the following steps: receiving the first drilling data, including field drilling data, logging while drilling (LWD), and data indicating wear of at least one part of at least one drilling tool; training at least one predictive bit wear model to predict at least one or more wear states of the drilling tool based, at least in part, on the first drilling data, where the predictive bit wear model is configured to encode the first drilling data into a latent space representation of the first drilling data; and establishing one or more similarity scores that respondTo compare the latent space representation of the initial drilling data with the target drilling data; to filter the initial drilling data or the latent space representation of the initial drilling data that responds to one or more similarity scores; to fit a wear trajectory regression curve to represent the filtered latent space of the initial drilling data; and to establish one or more predicted wear states for at least one drilling tool based, at least in part, on a predictive bit wear model and a wear trajectory regression curve.
