Wear detection method, device and equipment

Abstract

The invention relates to the technical field of petroleum and natural gas drilling construction optimization, in particular to a wear detection method, device and equipment. The wear detection method comprises the following steps: clustering first wear data of a drill bit; the first wear data comprises second wear data and third wear data; the second wear data represents historical wear data of the drill bit in the current well drilling process; the third wear data represents real-time wear data of the drill bit in the current well drilling process; feature migration is carried out in the clustering process of the first wear data based on fourth wear data of the drill bit; the fourth wear data represents normal historical wear data of the drill bit in the historical well drilling process; and determining the real-time wear level of the drill bit based on the loss value from the third wear data to the first wear data clustering center and the fourth wear data of the drill bit. According to the embodiment of the invention, the wear data of the drill bit is obtained, clustering of the wear data is accelerated by using feature migration, and the accuracy and real-time performance of wear detection are improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of oil and gas drilling construction optimization, and specifically to a wear detection method, device and equipment. Background Art

[0002] The rock properties of unconventional formations are dense, hard, and highly abrasive, which leads to severe wear of the drill bit during the drilling process. Complex drill bit working conditions such as drill bit wear and stick-slip vibration are common, resulting in low drilling speed, long drilling cycle, and high cost. Complex conditions such as high temperature and pressure, high rock hardness, strong abrasiveness, and strong heterogeneity are difficult to avoid, which leads to limited understanding of drill bit working conditions and frequent abnormal rock breaking conditions, which have become an important factor restricting the speed and efficiency of deep and unconventional complex oil and gas drilling. Therefore, accurately identifying the working conditions of the drill bit in the well and realizing real-time monitoring of the rock breaking process are important bases and foundations for optimizing and controlling the rock breaking process in oil and gas drilling. Drill bit wear will directly affect the mechanical drilling speed during the drilling process, thereby affecting the progress and cost of the entire drilling project. However, the wear process of the drill bit in the well is hidden, and the well conditions cannot be known on site. The well conditions can only be indirectly understood through changes in ground parameters, and the drill bit status monitoring is difficult.

[0003] Existing drill bit wear detection methods mainly include traditional drill bit condition monitoring methods and drill bit wear condition prediction methods based on machine learning models. Traditional drill bit condition monitoring methods usually rely on comprehensive analysis of indoor tests and field data to determine the mapping relationship between the change trend of parameters such as mechanical penetration rate and mechanical specific energy and the drill bit condition. The basic idea of ​​these methods is to infer the rock breaking state and wear condition of the drill bit by analyzing various parameters collected during the drilling process. Machine learning algorithms establish data-driven models to achieve real-time monitoring and prediction of drill bit wear conditions. Compared with traditional physical model-based methods, machine learning technology can significantly improve the accuracy and real-time performance of monitoring, thereby providing more reliable decision support for drilling operations. The drill bit wear condition prediction method based on machine learning models first collects and stores the data generated during the drilling process, and then constructs a suitable training data set through feature extraction. Then, machine learning algorithms (such as support vector machines, random forests, decision trees, and logistic regression) are applied to train the data to generate a model that can identify and predict the drill bit wear condition. Finally, through a real-time monitoring system, the model is applied to actual drilling operations to achieve dynamic evaluation of drill bit wear.

[0004] The traditional drill bit status monitoring method judges the working status of the downhole drill bit based on the mechanical drilling speed, but the mechanical drilling speed is affected by many factors, and it is difficult to determine whether the reduction in drilling speed is caused by the reduction in the cutting performance of the drill bit itself. At the same time, there is no point-to-point mapping relationship between the mechanical drilling speed and the drill bit wear (i.e., low drilling speed does not mean high drill bit wear, and high drilling speed does not necessarily mean low drill bit wear), which leads to an incomplete evaluation of the comprehensive performance of the drill bit, reducing the accuracy and reliability of monitoring. In addition, in actual drilling operations, the response speed of the drill bit wear status prediction method based on the machine learning model often cannot meet the needs of real-time monitoring, and the drill bit wear label is difficult to obtain. The drill bit wear status can only be obtained when the drill bit is out of the well, resulting in the accuracy and real-time performance of the existing drill bit wear status prediction method based on the machine learning model to be improved. Therefore, how to overcome the problems of incomplete evaluation factors, difficulty in obtaining drill bit wear labels, poor real-time performance, and low accuracy in the existing methods, and propose a drill bit wear detection method with comprehensive evaluation factors, strong real-time performance, and high accuracy is a key problem that needs to be solved urgently. Summary of the invention

[0005] The purpose of the embodiments of this specification is to provide a wear detection method, device and equipment to overcome the problems existing in the existing methods, such as incomplete evaluation factors, difficulty in obtaining drill wear labels, poor real-time performance and low accuracy, and to propose a drill wear detection method with comprehensive evaluation factors, strong real-time performance and high accuracy.

[0006] On the one hand, an embodiment of the present specification proposes a wear detection method, which includes: clustering first wear data of a drill bit; the first wear data includes second wear data and third wear data; the second wear data represents historical wear data of the drill bit during the drilling of a current well; the third wear data represents real-time wear data of the drill bit during the drilling of the current well; based on the fourth wear data of the drill bit, performing feature migration in the clustering process of the first wear data; the fourth wear data represents normal historical wear data of the drill bit during the drilling of historical wells; based on the loss value from the third wear data to the cluster center of the first wear data and the fourth wear data of the drill bit, determining the real-time wear level of the drill bit.

[0007] On the other hand, a wear detection device comprises: a clustering module for clustering first wear data of a drill bit; the first wear data comprises second wear data and third wear data; the second wear data represents historical wear data of the drill bit during drilling of a current well; the third wear data represents real-time wear data of the drill bit during drilling of the current well; a migration module for performing feature migration in the clustering process of the first wear data based on fourth wear data of the drill bit; the fourth wear data represents normal historical wear data of the drill bit during drilling of historical wells; a determination module for determining the real-time wear level of the drill bit based on the loss value from the third wear data to the cluster center of the first wear data and the fourth wear data of the drill bit.

[0008] On the other hand, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the wear detection method.

[0009] It can be seen from the technical solutions provided in the above embodiments of this specification that the embodiments of this specification can cluster the first wear data of the drill bit; the first wear data includes the second wear data and the third wear data; the second wear data represents the historical wear data of the drill bit during the drilling of the current well; the third wear data represents the real-time wear data of the drill bit during the drilling of the current well; based on the fourth wear data of the drill bit, feature migration is performed in the clustering process of the first wear data; the fourth wear data represents the normal historical wear data of the drill bit during the drilling of the historical well; based on the loss value from the third wear data to the cluster center of the first wear data and the fourth wear data of the drill bit, the real-time wear level of the drill bit is determined. Compared with the existing method, the embodiments of this specification can obtain the wear data of the drill bit, and cluster the wear data of the drill bit to comprehensively evaluate the wear status of the drill bit. In addition, the use of feature migration can speed up the clustering process of the drill bit wear data, and improve the accuracy and real-time performance of the drill bit wear detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the drawings required for use in the embodiments or the prior art description are briefly introduced below.

[0011] Figure 1 is a flow chart of the wear detection method provided in the embodiment of this specification;

[0012] Figure 2 is an overall logic flow chart of the wear detection method provided in the embodiments of this specification;

[0013] Figure 3 It is a schematic diagram of an unknown wear process when the drill bit is not pulled out during a drilling operation provided in an embodiment of this specification;

[0014] Figure 4 is a schematic diagram of the change in the wear level of the drill bit provided in the embodiment of this specification;

[0015] Figure 5 is a flowchart of processing normal historical wear data and abnormal historical wear data provided by the embodiments of this specification;

[0016] Figure 6 It is a schematic diagram of the structure of the wear detection device provided in the embodiment of this specification;

[0017] Figure 7 It is a schematic diagram of the structural composition of a computer device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0019] Figure 1 is a flow chart of the wear detection method, Figure 2 The overall logic flow chart of the wear detection method is shown in Figure 2. Figure 1 and Figure 2 , when implemented specifically, the method comprises the following steps:

[0020] S101: Clustering the first wear data of the drill bit; the first wear data includes second wear data and third wear data; the second wear data represents the historical wear data of the drill bit during the drilling of the current well; the third wear data represents the real-time wear data of the drill bit during the drilling of the current well.

[0021] In some embodiments, drilling parameters of the drill bit can be obtained; based on correlation analysis, wear parameters in the drilling parameters can be screened; the wear parameters can be encoded using a pre-trained encoder to generate encoding parameters; based on the encoding parameters, fourth wear data of the drill bit can be obtained.

[0022] Through correlation analysis, wear parameters that are strongly correlated with drill bit wear can be comprehensively and effectively selected, improving the accuracy and reliability of the data and providing a comprehensive and high-quality data foundation for subsequent analysis. In addition, using a pre-trained encoder to encode wear parameters can reduce the dimension of the data, which is conducive to improving the real-time performance of wear detection.

[0023] The logging data and well logging data of historical wells can be obtained. Logging data is the data obtained by observing and recording the materials (such as cuttings, drilling fluid, gas, etc.) returned from the well and information such as drilling parameters during the drilling process. Well logging data is usually collected after the wellbore is drilled, and the data is transmitted to the ground through logging instruments for recording and analysis. Logging data and well logging data contain many drilling parameters related to the drill bit. The logging data and well logging data of the historical wells obtained can be preprocessed, including outlier detection, outlier removal, data completion and interpolation, and the sliding filter algorithm can be used to reduce the noise of the data. The correlation coefficient of different drilling parameters on drill bit wear can be calculated through correlation analysis, and all drilling parameters can be screened based on the correlation coefficient to find the wear parameter most associated with drill bit wear. Correlation analysis can be based on statistical methods such as Pearson correlation coefficient, Spearman correlation coefficient and determination coefficient. For example, the Pearson correlation coefficient can be used to calculate the correlation between each drilling parameter and drill bit wear, and the correlation strength between each drilling parameter and drill bit wear can be determined based on the absolute value of the calculated correlation coefficient. A correlation strength threshold can be set, for example, the correlation strength threshold is 0.6. Among all drilling parameters, the drilling parameters with a correlation strength greater than or equal to the correlation strength threshold are selected as wear parameters.

[0024] In some embodiments, the wear parameters obtained by screening using the Pearson correlation coefficient may at least include: drill bit size, cutting tooth diameter, number of blades, number of water holes, drilling pressure, rotation speed, torque, displacement, and pump pressure. Among them, the drill bit size has a large correlation with the type and purpose of the drill bit. The cutting tooth diameter and the number of blades have an important influence on the average drilling speed and rock breaking efficiency of the drill bit. The number of water holes is related to the cooling and chip removal effect of the drill bit. During the drilling process, the water holes can help transport the drilling fluid to the vicinity of the drill bit to cool the drill bit and remove the cuttings. The drilling pressure refers to the pressure generated by the gravity of part of the drill string added to the drill bit when the drill string is lowered. The magnitude of the drilling pressure has a great influence on the drilling efficiency and the life of the drill bit. The rotation speed refers to the rotation speed of the drill bit during the drilling process. The speed will affect the drilling efficiency and the wear of the drill bit. Torque refers to the torque generated by the drill bit when it rotates. The magnitude of the torque reflects the load capacity of the drill bit during the drilling process. The displacement refers to the flow rate of drilling fluid during the drilling process. The displacement will affect the cooling, cleaning and cuttings carrying effects of the drilling fluid. The pump pressure refers to the pressure generated when the drilling fluid is pumped to the vicinity of the drill bit during the drilling process. The pump pressure will affect the delivery effect of the drilling fluid and the cooling of the drill bit.

[0025] After the wear parameters are determined, a pre-trained encoder can be used to encode the wear parameters. The encoder can be a machine learning model such as RNN, LSTM or GRU. For example, a pre-trained GRU can be used as an encoder. First, the input wear parameter sequence can be converted into a wear vector sequence using the embedding of the encoder, and then the wear vector sequence is sent to the encoding layer. The encoder can be used to process the wear vector sequence one by one, and its hidden state is updated each time it is processed. The final hidden state is used as the encoding parameter representation of the entire wear vector sequence. Based on the normal logging data and normal logging data of historical wells, normal wear parameter data of multiple historical wells at different locations can be obtained. Use a pre-trained encoder to encode the normal wear parameter data of each historical well at each time point to obtain the encoding parameter representation corresponding to each historical well at each time point. Based on the encoding parameter representation of all positions of all historical wells, the fourth wear data used to characterize the normal wear of historical wells can be obtained.

[0026] In some embodiments, based on the internal and external tooth wear ratings of the drill bit corresponding to the fourth wear data, the number of cluster centers of the fourth wear data can be determined; based on the number of cluster centers of the fourth wear data, the fourth wear data can be clustered; based on the cluster centers of the fourth wear data, a wear level label for the fourth wear data can be generated; and the cluster centers of the fourth wear data can be recorded to generate a historical cluster center set.

[0027] By generating wear grade labels for normal historical wear data, the amount of drill wear labels is expanded, solving the problem that drill wear labels are difficult to obtain, and providing sufficient training sample data for wear detection, which helps to improve the accuracy of drill wear detection. In addition, the cluster center of the fourth wear data is recorded to generate a historical cluster center set, laying a data foundation for feature migration.

[0028] Reference Figure 3 and Figure 4, the wear level of the drill bit cannot be obtained in real-time drilling, that is, before the drill bit leaves the well. The wear level of the drill bit can be determined by the drill bit wear rating method after the drill bit leaves the well. The drill bit wear rating method can be, for example, the internal and external tooth wear IADC drill bit wear grading standard method. The internal and external tooth wear IADC drill bit wear grading standard method describes the cutting structure, wear characteristics, position, bearing / seal status, gauge wear and other related information of the drill bit in detail through specific coding and grading. Among them, the inner teeth of the drill bit represent the inner 2 / 3 of the drill bit radius in a fixed cutting tooth drill bit. The inner teeth are mainly responsible for breaking rocks during the drilling process. The outer teeth of the drill bit represent the outer 1 / 3 of the drill bit radius in a fixed cutting tooth drill bit. The outer teeth also play a role in breaking rocks during the drilling process, but their wear conditions are usually different from those of the inner teeth. The internal and external tooth wear IADC drill bit wear grading standard method uses linear numbers from 0 to 8 to measure and define the wear status of the cutting teeth at various parts of the drill bit surface. The larger the value of the number, the greater the wear of the cutting teeth. 0 means no wear on the cutting teeth, 4 means 50% wear on the cutting teeth, and 8 means the cutting teeth are completely worn out with nothing left. The IADC drill wear grading standard method calculates the average wear level of each tooth in the inner and outer tooth areas to obtain the average wear level of each area, and then determines the wear level of the inner and outer teeth.

[0029] Therefore, for the fourth wear data related to multiple historical wells, the internal and external tooth wear IADC drill bit wear grading standard method can be used to obtain the internal tooth wear level and external tooth wear level of the corresponding drill bit after the drilling of each historical well. The internal tooth wear level and the external tooth wear level of the drill bit can be fused to obtain a wear level that can characterize the overall wear of the drill bit. The fusion can be a simple arithmetic average of the internal tooth wear level and the external tooth wear level, or a more complex weighted superposition. For example, the arithmetic average of the internal tooth wear level and the external tooth wear level of the drill bit can be taken to generate a wear level that characterizes the overall wear of the drill bit. For another example, different weight factors can be assigned to the internal tooth wear level and the external tooth wear level of the drill bit according to the actual role of the internal teeth and external teeth of the drill bit in the drilling process, and then the wear level that can characterize the overall wear of the drill bit can be calculated.

[0030] Reference Figure 5After determining the wear level of the drill bit corresponding to each historical well, the maximum wear level can be taken as the number of cluster centers corresponding to the fourth wear data. According to the number of cluster centers corresponding to the fourth wear data, the fourth wear data can be clustered. For example, according to the number of cluster centers corresponding to the fourth wear data, the fourth wear data can be clustered using the K-Means clustering algorithm. K fourth wear data points can be randomly selected as the initial cluster centers. For each data point in the fourth wear data, the distance between it and each cluster center is calculated, and it is assigned to the cluster center with the closest distance. Then, for each cluster, its cluster center is recalculated. The new cluster center is the mean of all fourth wear data points in the cluster. After multiple rounds of iterations, until the cluster center no longer changes. For each cluster center of the fourth wear data finally obtained, each cluster center can represent a wear level. Therefore, for each fourth wear data point, the wear level represented by the cluster center closest to it can be used as its wear level label. For each cluster center of the fourth wear data finally obtained, the position of each cluster center can also be recorded to generate a set of historical cluster centers.

[0031] In some embodiments, based on the coding parameters, the fifth wear data of the drill bit can be obtained; the fifth wear data represents the abnormal historical wear data of the drill bit during the historical well drilling process; the abnormal operating conditions corresponding to the fifth wear data can be determined; based on the fifth wear data and the fourth wear data, the confidence interval of the wear parameter can be determined.

[0032] By analyzing the abnormal wear process and carrying out corresponding marking and processing, the foundation is laid for timely identifying abnormal conditions of the drill bit under abnormal working conditions during real-time wear detection, which in turn helps to enhance the safety of actual drilling operations.

[0033] After the wear parameters are determined, a pre-trained encoder can be used to encode the wear parameters. The encoder can be a machine learning model such as RNN, LSTM or GRU. For example, a pre-trained GRU can be used as an encoder. First, the input wear parameter sequence can be converted into a wear vector sequence using the embedding of the encoder, and then the wear vector sequence is sent to the encoding layer. The encoder can be used to process the wear vector sequence one by one, and its hidden state is updated each time it is processed. The final hidden state is used as the encoding parameter representation of the entire wear vector sequence. Based on the abnormal logging data and abnormal logging data of historical wells, abnormal wear parameter data of multiple historical wells at different locations can be obtained. Use a pre-trained encoder to encode the abnormal wear parameter data of each historical well at each time point to obtain the encoding parameter representation corresponding to each historical well at each time point. Based on the encoding parameter representation of all positions of all historical wells, the fifth wear data for characterizing the abnormal wear of historical wells can be obtained.

[0034] For the fifth wear data related to multiple historical wells, the internal and external tooth wear IADC drill bit wear grading standard method can be used to obtain the internal and external tooth wear levels of the corresponding drill bit after the drilling of each historical well. The internal and external tooth wear levels of the drill bit can be fused to obtain a wear level that can characterize the overall wear of the drill bit. The fusion can be a simple arithmetic average of the internal and external tooth wear levels, or a more complex weighted superposition. For example, the arithmetic average of the internal and external tooth wear levels of the drill bit can be taken to generate a wear level that characterizes the overall wear of the drill bit. For another example, different weight factors can be assigned to the internal and external tooth wear levels of the drill bit according to the actual role of the internal and external teeth of the drill bit in the drilling process, and then the wear level that can characterize the overall wear of the drill bit can be calculated. After determining the wear level of the drill bit corresponding to each historical well, the maximum wear level can be taken as the number of cluster centers corresponding to the fifth wear data. According to the number of cluster centers corresponding to the fifth wear data, the fifth wear data can be clustered using the K-Means clustering algorithm. K fifth wear data points can be randomly selected as the initial cluster centers. For each data point in the fifth wear data, its distance to each cluster center is calculated and assigned to the cluster center with the nearest distance. Then, for each cluster, its cluster center is recalculated. The new cluster center is the mean of all fourth wear data points in the cluster. After multiple rounds of iterations, until the cluster center no longer changes. For each cluster center of the fifth wear data finally obtained, the distance loss value between different cluster centers is calculated. Based on the clustering results of the fourth wear data, a set of historical cluster centers can be obtained. The distance loss value between each cluster center in the set of historical cluster centers can be calculated. It can be found that the distance loss value between each cluster center of the fifth wear data is much larger than the distance loss value between each cluster center of the fourth wear data. This is because some wear parameters related to the fifth wear data have mutated. According to the abnormal logging data and abnormal logging data of historical wells, the abnormal working conditions corresponding to each fifth wear data can be determined. Abnormal working conditions may include but are not limited to abnormal wear conditions such as broken teeth, broken teeth, mud balls, etc. Reference Figure 5According to the wear parameter variation range of the fourth wear data and the wear parameter variation range of the fifth wear data, a digital signal analysis method can be used to mark and determine the fluctuation range of the wear parameter variation corresponding to each abnormal working condition and the confidence interval of the wear parameter variation under normal wear. The digital signal analysis method can be a short-time Fourier transform, a wavelet transform, an HHT transform, etc. For example, a wavelet transform analysis mark can be used to determine the fluctuation range of the wear parameter variation corresponding to each abnormal working condition and the confidence interval of the wear parameter variation under normal wear. Specifically, the wear parameter data corresponding to the plurality of fourth wear data and the plurality of fifth wear data can be decomposed by wavelet transform to obtain sub-wear signals of different frequencies. By comparing the features of adjacent sub-wear signals (such as energy, amplitude, frequency, etc.), the position of the mutation point can be detected. For the mutation point, the position of the mutation point can be accurately located by interpolation or other methods. At the mutation point, relevant parameters such as amplitude, frequency, phase, etc. are extracted. According to the extracted parameters, the variation range of the mutation parameter is determined, and then the fluctuation range of the wear parameter variation corresponding to each abnormal working condition and the confidence interval of the wear parameter variation under normal wear can be determined.

[0035] In some embodiments, the first wear data may be clustered for different numbers of cluster centers; based on the historical cluster center set of the fourth wear data, the cluster centers of the first wear data with different numbers of cluster centers may be selected.

[0036] By optimizing the random selection of cluster centers to a method based on a set of historical cluster centers, the clustering iteration speed of the first wear data is accelerated, which helps to improve the real-time response capability of wear detection.

[0037] The IADC drill wear classification standard method for internal and external tooth wear uses linear numbers from 0 to 8 to measure and define the wear state of the cutting teeth at various parts of the drill surface. Therefore, the number of cluster centers of the first wear data can be set to 1-9, and the first wear data can be clustered under 9 different numbers of cluster centers. In the process of clustering the first wear data, the cluster center of the first wear data can be selected from the historical cluster center set, thereby accelerating the convergence speed of the clustering algorithm. For example, according to the number of cluster centers corresponding to the first wear data, the K-Means clustering algorithm can be used to cluster the first wear data. K first wear data points can be randomly selected. For each selected first wear data point, the cluster center with the smallest distance to it can be selected from the historical cluster center set as the initial cluster center of the first wear data clustering process. For each data point in the first wear data, calculate its distance to each cluster center and assign it to the cluster center with the closest distance. Then, for each cluster, recalculate its cluster center. The new cluster center is the mean of all first wear data points in the cluster. After multiple rounds of iterations, the cluster center no longer changes.

[0038] S102: performing feature migration in the clustering process of the first wear data based on fourth wear data of the drill bit; the fourth wear data represents normal historical wear data of the drill bit during the historical well drilling process.

[0039] In some embodiments, the fourth wear data of the drill bit can be clustered; the first wear data clustering loss and the fourth wear data clustering loss can be obtained; based on minimizing the difference between the first wear data clustering loss and the fourth wear data clustering loss, the feature distribution in the fourth wear data clustering process can be learned in the first wear data clustering process.

[0040] By introducing the feature migration method, the learning speed in the first wear data clustering process can be accelerated, the training time can be reduced, and the real-time response capability of wear detection can be improved. Then, the drill bit wear can be monitored in real time, ensuring timely identification of the wear level during the drilling process, avoiding equipment damage and operation interruption caused by wear.

[0041] The number of cluster centers of the first wear data can be set to 1-9, and the first wear data can be clustered under 9 different numbers of cluster centers. For each clustering, the fourth wear data can be clustered at the same time. The fourth wear data has a corresponding wear level label, and the first wear data does not have a corresponding wear level label. The clustering loss of the first wear data and the clustering loss of the fourth wear data can be obtained, and the difference between the two clustering losses can be minimized through the transfer learning model, and then the feature distribution of the fourth wear data clustering process can be learned in the process of clustering the first wear data. The transfer learning model can be an MMD, TCA, DAN and other algorithms.

[0042] For example, MMD can be used to learn the feature distribution in the fourth wear data clustering process during the first wear data clustering process. The clustering loss of the first wear data and the clustering loss of the fourth wear data can be obtained, and the difference between the clustering loss of the first wear data and the clustering loss of the fourth wear data can be minimized based on the following formula, thereby learning the feature distribution in the fourth wear data clustering process during the first wear data clustering process.

[0043]

[0044] Loss = α1·λ1-α2·λ2

[0045] Among them, D s D represents the source domain with wear level labels, i.e., the fourth wear data; t is a target domain without a wear level label, i.e., the first wear data; represents the kernel function of MMD; λ1 and λ2 represent the clustering loss of the fourth wear data and the clustering loss of the first wear data, respectively; α1 and α2 represent the adjustable weight parameter of the clustering loss of the fourth wear data and the adjustable weight parameter of the clustering loss of the first wear data, respectively. The clustering process is a two-stage training process. In the first stage, α1 can be set to be greater than α2 until λ1 converges; in the second stage, α1 is gradually decayed to 0 until λ2 converges. Through the two-stage feature migration clustering, the feature distribution in the fourth wear data clustering process can be learned in the first wear data clustering process.

[0046] S103: Determine a real-time wear level of the drill bit based on the loss value from the third wear data to the cluster center of the first wear data and fourth wear data of the drill bit.

[0047] In some embodiments, the loss value from the third wear data to the cluster center of the first wear data may be calculated; based on the loss value corresponding to the third wear data under different numbers of cluster centers, the wear change amount of the third wear data may be calculated.

[0048] By analyzing the relationship between the cluster center and the third wear data, the interpretability of the wear detection method is enhanced, allowing users to understand the decision logic of the model and increase the trust in field applications. In addition, by calculating the wear change corresponding to the third wear data, a data foundation is laid for taking corresponding countermeasures based on normal wear and abnormal wear conditions.

[0049] The number of cluster centers of the first wear data can be set to 1-9, and the first wear data can be clustered under 9 different numbers of cluster centers. The clustering results of the first wear data under 9 different numbers of cluster centers can be obtained. For each case, the loss value between the third wear data and each cluster center of the first wear data is calculated, and the minimum loss value is taken. The corresponding loss values ​​in the 9 cases can be averaged by the algorithm, or the minimum value can be taken, or other methods can be used to calculate the loss value of the fused third wear data, which will not be repeated here. The loss value of the third wear data recorded at the previous time point can be obtained, and the difference between the loss value of the third wear data at the current time point and the loss value of the third wear data at the previous time point is taken as the wear change amount.

[0050] In some embodiments, if the wear change is less than a preset wear change threshold, the number of cluster centers of the first wear data can be determined based on the loss value corresponding to the third wear data under different numbers of cluster centers and the fourth wear data; based on the number of cluster centers of the first wear data, the real-time wear level of the drill bit can be determined.

[0051] When the wear change is less than a preset wear change threshold, the accurate real-time wear level of the drill bit can be obtained based on the number of cluster centers of the first wear data, thereby achieving real-time monitoring of the wear condition of the drill bit.

[0052] Based on the clustering result of the fourth wear data, a wear change threshold can be preset. Specifically, in the clustering process of the fourth wear data, the difference between the corresponding loss values ​​of the fourth wear data at adjacent time points can be calculated. The maximum value of the difference obtained by multiple calculations can be taken, and an amplification factor greater than 1 can be set, and the product of the maximum value of the difference obtained by multiple calculations and the amplification factor is used as the wear change threshold. If the wear change amount of the third wear data is less than the preset wear change threshold, it can be considered that the third wear data at this time is normal wear data. Similarly, based on the clustering result of the fourth wear data, a historical cluster center set can be obtained. Each cluster center in the historical cluster center set can represent a wear level, and based on the fourth wear data corresponding to each cluster center in the historical cluster center set, the loss value fluctuation range corresponding to each wear level can be determined. For the clustering results of the first wear data under 9 different numbers of cluster centers, for each case, the loss value between the third wear data and each cluster center of the first wear data is calculated, and the minimum loss value is taken. Obtain the fluctuation range of the loss value corresponding to the wear level under each number of cluster centers, and compare it with the minimum loss value corresponding to the third wear data under each number of cluster centers, and determine the number of cluster centers whose minimum loss value corresponding to the third wear data is most matched with the fluctuation range of the loss value corresponding to the wear level. Take this number of cluster centers as the number of cluster centers of the first wear data. Under normal wear conditions, the wear level of the drill bit increases approximately linearly with the deepening of the drilling depth, so the cluster center of the current first wear data can represent the wear level corresponding to the lowest wear level to the current third wear data in sequence. The cluster center corresponding to the third wear data at the current time point corresponds to the current maximum wear level, that is, the number of cluster centers of the first wear data is the real-time wear level corresponding to the third wear data at the current time point. For example, based on the IADC drill wear grading standard method for internal and external tooth wear, the number of cluster centers of the current first wear data is 5, which indicates that the cluster center of the first wear data can represent the 0th wear level, the 1st wear level, the 2nd wear level, the 3rd wear level and the 4th wear level in sequence. The real-time wear level corresponding to the third wear data is the number of cluster centers of the current first wear data, that is, the real-time wear level corresponding to the third wear data is the fourth wear level.

[0053] In some embodiments, if the wear change is greater than or equal to a preset wear change threshold, a pre-trained decoder can be used to decode the third wear data to generate wear parameter data; based on the wear parameter data and the confidence interval of the wear parameter, the abnormal operating condition corresponding to the third wear data can be determined.

[0054] When the wear change is greater than or equal to the preset wear change threshold, the abnormal operating condition corresponding to the third wear data can be determined based on the wear parameter data and the confidence interval of the wear parameter, and then corresponding countermeasures for the abnormal operating condition can be taken, which helps to ensure the safety of actual drilling operations.

[0055] Based on the clustering result of the fourth wear data, a wear change threshold can be preset. Specifically, in the clustering process of the fourth wear data, the difference between the corresponding loss values ​​of the fourth wear data at adjacent time points can be calculated. The maximum value of the difference values ​​calculated multiple times can be taken, and an amplification factor greater than 1 can be set, and the product of the maximum value of the difference values ​​calculated multiple times and the amplification factor is used as the wear change threshold. If the wear change amount of the third wear data is greater than or equal to the preset wear change threshold, it can be considered that the third wear data at this time is abnormal wear data. The third wear data can be decoded using a pre-trained decoder. The decoder can be a machine learning model such as RNN, LSTM or GRU. For example, a pre-trained GRU can be used as a decoder. The third wear data can be directly used as the output of the decoder, and the decoder decodes the third wear data and generates wear parameter data. According to the wear parameter change range of the fourth wear data and the wear parameter change range of the fifth wear data, the fluctuation range of the wear parameter change corresponding to each abnormal working condition and the confidence interval of the wear parameter change under normal wear conditions are determined. Each wear parameter in the wear parameter data corresponding to the third wear data can be matched with the fluctuation range of the wear parameter change corresponding to each abnormal operating condition and the confidence interval of the wear parameter change under normal wear conditions, so as to determine the abnormal operating condition corresponding to the third wear data.

[0056] In some embodiments, a real wear level label of the first wear data after it leaves the wellbore may be obtained; based on the real wear level label, the clustering result of the first wear data may be optimized.

[0057] By obtaining the real wear level label of the drill bit after leaving the well, the data can be corrected and the clustering results can be updated in real time to further improve the accuracy and reliability of the wear detection method.

[0058] After the drilling work of the current well is completed, the internal and external tooth wear IADC drill bit wear grading standard method can be used to obtain the internal and external tooth wear level of the drill bit after each current well is out of the well. The internal and external tooth wear levels of the drill bit can be fused to obtain a wear level that can characterize the overall wear of the drill bit. The fusion can be a simple arithmetic average of the internal and external tooth wear levels, or a more complex weighted superposition. For example, the arithmetic average of the internal and external tooth wear levels of the drill bit can be taken to generate a wear level that characterizes the overall wear of the drill bit. For another example, different weight factors can be assigned to the internal and external tooth wear levels of the drill bit according to the actual role of the internal and external teeth of the drill bit in the drilling process, and then the real wear level label that can characterize the overall wear of the drill bit can be calculated.

[0059] After determining the true wear level label of the current well drill bit, the true wear level label of the current well drill bit can be used as the number of cluster centers of the first wear data. According to the number of cluster centers corresponding to the first wear data, the first wear data can be clustered. For example, according to the number of cluster centers corresponding to the first wear data, the K-Means clustering algorithm can be used to cluster the first wear data. K first wear data points can be randomly selected as the initial cluster centers. For each data point in the first wear data, its distance from each cluster center is calculated and assigned to the cluster center with the closest distance. Then, for each cluster, its cluster center is recalculated. The new cluster center is the mean of all the first wear data points in the cluster. After multiple rounds of iterations, until the cluster center no longer changes. For each cluster center of the first wear data finally obtained, each cluster center can represent a wear level. Therefore, for each first wear data point, the wear level represented by the cluster center closest to it can be used as its wear level label. For each cluster center of the first wear data finally obtained, the position of each cluster center can also be recorded, and the historical cluster center set can be updated, thereby updating the confidence interval of the wear parameter change under normal wear conditions, which will not be repeated here.

[0060] Based on the above wear detection method, this specification also proposes an embodiment of a wear detection device. Figure 6 As shown, the evaluation device may specifically include the following modules:

[0061] The clustering module 601 can be used to cluster the first wear data of the drill bit; the first wear data includes the second wear data and the third wear data; the second wear data represents the historical wear data of the drill bit during the drilling of the current well; the third wear data represents the real-time wear data of the drill bit during the drilling of the current well;

[0062] Migration module 602, which can be used to perform feature migration in the clustering process of the first wear data based on fourth wear data of the drill bit; the fourth wear data represents normal historical wear data of the drill bit during the historical well drilling process;

[0063] The determination module 603 may be configured to determine the real-time wear level of the drill bit based on the loss value from the third wear data to the cluster center of the first wear data and the fourth wear data of the drill bit.

[0064] In some embodiments, the clustering module 601 can be specifically used to obtain drilling parameters of the drill bit; based on Pearson correlation detection, screen wear parameters in the drilling parameters; encode the wear parameters using a pre-trained encoder to generate encoding parameters; based on the encoding parameters, obtain fourth wear data of the drill bit.

[0065] In some embodiments, the above-mentioned clustering module 601 can also be specifically used to determine the number of cluster centers of the fourth wear data based on the internal and external tooth wear ratings of the drill bit corresponding to the fourth wear data; cluster the fourth wear data based on the number of cluster centers of the fourth wear data; generate a wear level label for the fourth wear data based on the fourth wear data and the cluster centers of the fourth wear data; record the cluster centers of the fourth wear data to generate a historical cluster center set.

[0066] In some embodiments, the above-mentioned clustering module 601 can also be specifically used to obtain the fifth wear data of the drill bit based on the encoding parameters; the fifth wear data represents the abnormal historical wear data of the drill bit during the historical well drilling process; determine the abnormal operating conditions corresponding to the fifth wear data; based on the abnormal operating conditions corresponding to the fifth wear data, determine the confidence interval of the wear parameter.

[0067] In some embodiments, the clustering module 601 can be specifically used to cluster the first wear data according to different numbers of cluster centers; based on the historical cluster center set of the fourth wear data, the cluster centers of the first wear data with different numbers of cluster centers are selected.

[0068] In some embodiments, the above-mentioned migration module 602 can be specifically used to cluster the fourth wear data of the drill bit; obtain the first wear data clustering loss and the fourth wear data clustering loss; based on minimizing the difference between the first wear data clustering loss and the fourth wear data clustering loss, learn the characteristic distribution in the fourth wear data clustering process during the first wear data clustering process.

[0069] In some embodiments, the above-mentioned determination module 603 can be specifically used to calculate the loss value from the third wear data to the cluster center of the first wear data; calculate the wear change of the third wear data based on the loss value corresponding to the third wear data under different numbers of cluster centers; if the wear change is less than a preset wear change threshold, determine the number of cluster centers of the first wear data based on the loss value corresponding to the third wear data under different numbers of cluster centers and the fourth wear data; determine the real-time wear level of the drill bit based on the number of cluster centers of the first wear data; if the wear change is greater than or equal to the preset wear change threshold, use a pre-trained decoder to decode the third wear data to generate wear parameter data; based on the wear parameter data and the confidence interval of the wear parameter, determine the abnormal operating condition corresponding to the third wear data.

[0070] In some embodiments, the determination module 603 may be specifically used to obtain a real wear level label of the first wear data after it is taken out of the well; and based on the real wear level label, optimize the clustering result of the first wear data.

[0071] It should be noted that the units, devices or modules described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described separately by functions divided into various modules. Of course, when implementing this specification, the functions of each module can be implemented in the same or more software and / or hardware, or the modules that implement the same function can be implemented by a combination of multiple sub-modules or sub-units. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0072] As can be seen from the above, based on the wear detection device provided in the embodiment of this specification, the wear data of the drill bit can be obtained, and the wear data of the drill bit can be clustered to comprehensively evaluate the wear state of the drill bit. In addition, by using feature migration, the clustering process of the drill bit wear data can be accelerated, thereby improving the accuracy and real-time performance of the drill bit wear detection.

[0073] The embodiment of the present specification also provides a computer device for a wear detection method, including a processor and a memory for storing instructions executable by the processor, and the processor can perform the following steps according to the instructions when implemented: clustering the first wear data of the drill bit; the first wear data includes second wear data and third wear data; the second wear data represents the historical wear data of the drill bit during the drilling of the current well; the third wear data represents the real-time wear data of the drill bit during the drilling of the current well; based on the fourth wear data of the drill bit, feature migration is performed in the clustering process of the first wear data; the fourth wear data represents the normal historical wear data of the drill bit during the drilling of historical wells; based on the loss value from the third wear data to the cluster center of the first wear data and the fourth wear data of the drill bit, the real-time wear level of the drill bit is determined.

[0074] In order to complete the above instructions more accurately, refer to Figure 7 As shown, the embodiment of this specification also provides another specific electronic device, wherein the electronic device includes a network communication port 701, a processor 702 and a memory 703, and the above structures are connected through internal cables so that each structure can perform specific data interaction.

[0075] The processor 702 can be specifically used to cluster the first wear data of the drill bit; the first wear data includes the second wear data and the third wear data; the second wear data represents the historical wear data of the drill bit during the drilling of the current well; the third wear data represents the real-time wear data of the drill bit during the drilling of the current well; based on the fourth wear data of the drill bit, feature migration is performed in the clustering process of the first wear data; the fourth wear data represents the normal historical wear data of the drill bit during the drilling of historical wells; based on the loss value from the third wear data to the cluster center of the first wear data and the fourth wear data of the drill bit, the real-time wear level of the drill bit is determined.

[0076] The memory 703 may be specifically used to store corresponding instruction programs.

[0077] In this embodiment, the network communication port 701 can be a virtual port that is bound to different communication protocols so that different data can be sent or received. For example, the network communication port can be a port responsible for web data communication, a port responsible for FTP data communication, or a port responsible for email data communication. In addition, the network communication port can also be a physical communication interface or communication chip. For example, it can be a wireless mobile network communication chip, such as GSM, CDMA, etc.; it can also be a Wifi chip; it can also be a Bluetooth chip.

[0078] In this embodiment, the processor 702 may be implemented in any appropriate manner. For example, the processor may take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) executable by the (micro)processor, a logic gate, a switch, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, etc. This specification does not limit this.

[0079] In this embodiment, the memory 703 may include multiple levels. In a digital system, anything that can store binary data can be a memory; in an integrated circuit, a circuit with a storage function that has no physical form is also called a memory, such as RAM, FIFO, etc.; in a system, a storage device with a physical form is also called a memory, such as a memory stick, TF card, etc.

[0080] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0081] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0082] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0083] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0084] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A wear detection method, characterized in that: include: Clustering first wear data of the drill bit; the first wear data includes second wear data and third wear data; The second wear data represents the historical wear data of the drill bit during the drilling of the current well; the third wear data represents the real-time wear data of the drill bit during the drilling of the current well; Based on the fourth wear data of the drill bit, feature migration is performed in the clustering process of the first wear data; The fourth wear data represents normal historical wear data of the drill bit during historical well drilling; Based on the loss value from the third wear data to the cluster center of the first wear data and the fourth wear data of the drill bit, a real-time wear level of the drill bit is determined.

2. The method according to claim 1, characterized in that The method further comprises: Obtain drilling parameters of the drill bit; Based on Pearson correlation detection, screening wear parameters in the drilling parameters; encoding the wear parameters using a pre-trained encoder to generate encoding parameters; Based on the encoding parameters, fourth wear data of the drill bit is acquired.

3. The method according to claim 2, characterized in that The method further comprises: Determining the number of cluster centers of the fourth wear data based on the wear ratings of the internal and external teeth of the drill bit corresponding to the fourth wear data; clustering the fourth wear data based on the number of cluster centers of the fourth wear data; generating a wear level label of the fourth wear data based on the fourth wear data and a cluster center of the fourth wear data; The cluster center of the fourth wear data is recorded to generate a historical cluster center set.

4. The method according to claim 2, characterized in that: The method further comprises: Based on the encoding parameters, fifth wear data of the drill bit is acquired; the fifth wear data represents abnormal historical wear data of the drill bit during the historical well drilling process; determining an abnormal operating condition corresponding to the fifth wear data; Based on the abnormal operating condition corresponding to the fifth wear data, a confidence interval of the wear parameter is determined.

5. The method according to claim 1, characterized in that The clustering of the first wear data of the drill bit comprises: According to different numbers of cluster centers, the first wear data are clustered respectively; Based on the historical cluster center set of the fourth wear data, cluster centers of the first wear data with different numbers of cluster centers are selected.

6. The method according to claim 1, characterized in that The fourth wear data based on the drill bit, performing feature migration in the clustering process of the first wear data, comprises: Clustering the fourth wear data of the drill bit; Obtaining the first wear data clustering loss and the fourth wear data clustering loss; Based on minimization of the difference between the first wear data clustering loss and the fourth wear data clustering loss, feature distribution in the fourth wear data clustering process is learned in the first wear data clustering process.

7. The method according to claim 1, characterized in that The determining the real-time wear level of the drill bit based on the loss value from the third wear data to the cluster center of the first wear data and the fourth wear data of the drill bit comprises: Calculating a loss value from the third wear data to the cluster center of the first wear data; Calculating the wear variation of the third wear data based on the loss values ​​corresponding to the third wear data under different numbers of cluster centers; If the wear change amount is less than a preset wear change threshold, determining the number of cluster centers of the first wear data based on the loss value corresponding to the third wear data under different numbers of cluster centers and the fourth wear data; Determining a real-time wear level of the drill bit based on the number of cluster centers of the first wear data; If the wear change amount is greater than or equal to a preset wear change threshold, use a pre-trained decoder to decode the third wear data to generate wear parameter data; Based on the wear parameter data and the confidence interval of the wear parameter, an abnormal operating condition corresponding to the third wear data is determined.

8. The method according to claim 1, characterized in that The method further comprises: Obtain the actual wear level label after the first wear data is taken out of the well; Based on the real wear level label, the clustering result of the first wear data is optimized.

9. A wear detection device, characterized in that: The device comprises: A clustering module, for clustering first wear data of the drill bit; the first wear data includes second wear data and third wear data; the second wear data represents historical wear data of the drill bit during the drilling of the current well; the third wear data represents real-time wear data of the drill bit during the drilling of the current well; A migration module, configured to perform feature migration in the clustering process of the first wear data based on fourth wear data of the drill bit; the fourth wear data represents normal historical wear data of the drill bit during the historical well drilling process; The determination module is used to determine the real-time wear level of the drill bit based on the loss value from the third wear data to the cluster center of the first wear data and the fourth wear data of the drill bit.

10. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 8 is implemented.

Images