Sucker rod cleaning and repair method

Abstract

The present application discloses a method for cleaning and repairing a sucker rod. The method comprises: grabbing a sucker rod to be cleaned; using a fully automatic closed high-temperature cleaning machine to remove rust, scale and wax from the sucker rod to be cleaned to obtain a cleaned sucker rod; transferring the cleaned sucker rod to a flaw detection line to obtain a flaw detection result; in response to the flaw detection result being unqualified, marking the cleaned sucker rod as unqualified; in response to the flaw detection result being qualified, repairing the cleaned sucker rod to obtain a repaired sucker rod. In the above method, a fully automatic closed high-temperature cleaning machine is used to efficiently and automatically remove rust, scale and wax from the sucker rod to ensure the cleaning effect and intelligence level.

Description

Technical Field

[0001] The present application relates to the field of sucker rod cleaning, and in particular to a sucker rod cleaning and repairing method. Background Art

[0002] The sucker rod is one of the key equipment in the oilfield exploitation process. Its main function is to pump the underground crude oil to the surface. Due to long-term work underground, the sucker rod will be affected by factors such as corrosion and wear, resulting in defects on its surface and internal structure. These defects will not only affect the service life of the sucker rod, but may also cause safety hazards in oil well operations. Therefore, regular cleaning and repair of the sucker rod is an important measure to ensure normal production and safe production in the oilfield.

[0003] However, the traditional sucker rod cleaning and repair methods have a low level of intelligence, and many cleaning and repair steps rely on manual operations, which is not only inefficient, but may also lead to inconsistent quality due to differences in operator skills. Although automated equipment may be used in some links, the overall process is not highly automated and intelligent, and many decisions and adjustments still require manual judgment. For example, when judging the damage of sucker rods, traditional methods may lack an intelligent diagnostic system and cannot quickly and accurately identify and evaluate the type and extent of damage, resulting in a large amount of manpower and material resources being wasted at the back end when repairing sucker rods, and the repair quality of sucker rods is unstable.

[0004] Therefore, an optimized sucker rod cleaning and repair solution is desired. Summary of the invention

[0005] This application is made in view of the above problems. One object of this application is to provide a method for cleaning and repairing a sucker rod.

[0006] An embodiment of the present application provides a method for cleaning and repairing a sucker rod, which includes: grabbing a sucker rod to be cleaned; using a fully automatic closed high-temperature cleaning machine to remove rust, scale and wax from the sucker rod to be cleaned to obtain a cleaned sucker rod; transferring the cleaned sucker rod to a flaw detection line to obtain a flaw detection result; in response to the flaw detection result being unqualified, marking the cleaned sucker rod as unqualified; in response to the flaw detection result being qualified, repairing the cleaned sucker rod to obtain a repaired sucker rod.

[0007] For example, according to the sucker rod cleaning and repair method of the embodiment of the present application, the cleaned sucker rod is transferred to the flaw detection line to obtain the flaw detection result, including: obtaining the leakage magnetic detection signal of the cleaned sucker rod; performing local signal segment feature extraction on the leakage magnetic detection signal to obtain a sequence of leakage magnetic detection signal segment current semantic association feature vectors; inputting the sequence of leakage magnetic detection signal segment current semantic association feature vectors into a self-supervised feature sequence significance fusion network to obtain a global leakage magnetic detection signal self-supervised significant fusion representation vector; based on the global leakage magnetic detection signal self-supervised significant fusion representation vector, determining whether there is an internal structural defect.

[0008] For example, according to the sucker rod cleaning and repair method of the embodiment of the present application, wherein local signal segment feature extraction is performed on the leakage magnetic detection signal to obtain a sequence of current semantic association feature vectors of the leakage magnetic detection signal segments, including: signal segmentation of the leakage magnetic detection signal to obtain a sequence of leakage magnetic detection signal segments; passing the sequence of leakage magnetic detection signal segments through a signal feature extractor based on a one-dimensional convolutional neural network model to obtain a sequence of current semantic association feature vectors of the leakage magnetic detection signal segments.

[0009] For example, according to the sucker rod cleaning and repair method of the embodiment of the present application, the sequence of the semantically associated feature vectors of the leakage magnetic detection signal segment current is input into the self-supervised feature sequence significance fusion network to obtain the global leakage magnetic detection signal self-supervised significant fusion representation vector, including: calculating the mean vector of the sequence of the semantically associated feature vectors of the leakage magnetic detection signal segment current as the center of the semantic distribution cluster of the leakage magnetic detection signal segment current to obtain the center feature vector of the semantic distribution cluster of the leakage magnetic detection signal segment current; calculating the semantically associated feature vectors of each leakage magnetic detection signal segment current in the sequence of the semantically associated feature vectors of the leakage magnetic detection signal segment current and the leakage magnetic detection signal. The semantic similarity score values ​​between the center feature vectors of the segment current semantic distribution clusters are obtained to obtain a sequence of similarity score values ​​of the center of the leakage magnetic detection signal segment current semantic distribution clusters; the sequence of similarity score values ​​of the leakage magnetic detection signal segment current semantic distribution cluster centers are normalized to obtain a sequence of similarity attention weight coefficients of the leakage magnetic detection signal segment current semantic distribution cluster centers; the sequence of similarity attention weight coefficients of the leakage magnetic detection signal segment current semantic distribution cluster centers are used as weights to calculate the vector-by-vector weighted sum of the sequence of the leakage magnetic detection signal segment current semantic association feature vectors to obtain the global leakage magnetic detection signal self-supervised significant fusion representation vector.

[0010] For example, according to the sucker rod cleaning and repair method of the embodiment of the present application, the semantic similarity score values ​​between each leakage magnetic detection signal segment current semantic association feature vector in the sequence of the leakage magnetic detection signal segment current semantic association feature vector and the central feature vector of the leakage magnetic detection signal segment current semantic distribution cluster are calculated to obtain a sequence of central similarity score values ​​of the leakage magnetic detection signal segment current semantic distribution cluster, including: respectively calculating the matrix product between each weight coefficient matrix and the corresponding leakage magnetic detection signal segment current semantic association feature vector to obtain a sequence of leakage magnetic detection signal segment current semantic weighted feature vectors; adding the corresponding bias vector to each leakage magnetic detection signal segment current semantic weighted feature vector in the sequence of the leakage magnetic detection signal segment current semantic weighted feature vector to obtain a sequence of leakage magnetic detection signal segment current semantic weighted bias adjusted feature vectors; inputting the sequence of leakage magnetic detection signal segment current semantic weighted bias adjusted feature vectors into the matrix product between each weight coefficient matrix and the corresponding leakage magnetic detection signal segment current semantic association feature vector to obtain a sequence of leakage magnetic detection signal segment current semantic weighted bias adjusted feature vectors. The function performs activation processing to obtain a sequence of magnetic flux leakage detection signal segment current semantic activation feature vectors; the product of each magnetic flux leakage detection signal segment current semantic activation feature vector in the sequence of magnetic flux leakage detection signal segment current semantic activation feature vectors and the central feature vector of the magnetic flux leakage detection signal segment current semantic distribution cluster is calculated respectively to obtain a sequence of similarity score values ​​of the central similarity of the magnetic flux leakage detection signal segment current semantic distribution cluster.

[0011] For example, according to the sucker rod cleaning and repair method of the embodiment of the present application, the sequence of the similarity score values ​​of the current semantic distribution cluster centers of the leakage magnetic detection signal segment is normalized to obtain a sequence of similarity attention weight coefficients of the current semantic distribution cluster centers of the leakage magnetic detection signal segment, including: masking the respective similarity score values ​​of the current semantic distribution cluster centers of the leakage magnetic detection signal segment in the sequence of the current semantic distribution cluster centers of the leakage magnetic detection signal segment to obtain a sequence of masked similarity score values ​​of the current semantic distribution cluster centers of the leakage magnetic detection signal segment; masking the respective masked similarity score values ​​of the current semantic distribution cluster centers of the masked leakage magnetic detection signal segment in the sequence of the current semantic distribution cluster centers of the leakage magnetic detection signal segment through A soft maximum normalization process is performed in the function to obtain a sequence of similarity attention weight coefficients of the center of the current semantic distribution cluster of the leakage magnetic detection signal segment.

[0012] For example, according to the sucker rod cleaning and repair method of the embodiment of the present application, each leakage magnetic detection signal segment current semantic distribution cluster center similarity score value in the sequence of leakage magnetic detection signal segment current semantic distribution cluster center similarity score values ​​is masked to obtain a sequence of masked leakage magnetic detection signal segment current semantic distribution cluster center similarity score values, including: in response to each leakage magnetic detection signal segment current semantic distribution cluster center similarity score value in the sequence of leakage magnetic detection signal segment current semantic distribution cluster center similarity score value being greater than or equal to a predetermined threshold, each leakage magnetic detection signal segment current semantic distribution cluster center similarity score value in the sequence of leakage magnetic detection signal segment current semantic distribution cluster center similarity score value is taken as the original value, otherwise it is set to zero.

[0013] For example, according to the sucker rod cleaning and repair method of the embodiment of the present application, based on the self-supervised significant fusion representation vector of the global magnetic leakage detection signal, determining whether there is an internal structural defect includes: passing the self-supervised significant fusion representation vector of the global magnetic leakage detection signal through a classifier-based flaw detection result generator to obtain a flaw detection result, and the flaw detection result is used to indicate whether there is an internal structural defect.

[0014] For example, according to the sucker rod cleaning and repair method of the embodiment of the present application, it also includes a training step: for training the signal feature extractor based on the one-dimensional convolutional neural network model, the self-supervised feature sequence significance fusion network and the classifier-based flaw detection result generator; wherein the training step includes: obtaining training data, the training data including the training leakage magnetic detection signal of the sucker rod after cleaning and the actual flaw detection result; performing signal segmentation on the training leakage magnetic detection signal to obtain a sequence of training leakage magnetic detection signal segments; passing the sequence of training leakage magnetic detection signal segments through the signal feature extractor based on the one-dimensional convolutional neural network model to obtain a sequence of current semantic association feature vectors of the training leakage magnetic detection signal segments; and performing current semantic association feature vectors on the training leakage magnetic detection signal segments. The sequence of feature vectors is input into the self-supervised feature sequence significance fusion network to obtain a training global magnetic flux leakage detection signal self-supervised significant fusion representation vector; the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector is optimized to obtain an optimized training global magnetic flux leakage detection signal self-supervised significant fusion representation vector; the optimized training global magnetic flux leakage detection signal self-supervised significant fusion representation vector is passed through the classifier-based flaw detection result generator to obtain a training flaw detection result; the cross entropy loss function value between the training flaw detection result and the true flaw detection result is calculated to obtain a classification loss function value; based on the classification loss function value, the signal feature extractor based on the one-dimensional convolutional neural network model, the self-supervised feature sequence significance fusion network and the classifier-based flaw detection result generator are trained.

[0015] According to the sucker rod cleaning and repair method of the embodiment of the present application, a fully automatic closed high-temperature cleaning machine is used to efficiently and automatically remove rust, scale and wax from the sucker rod, ensuring the cleaning effect and intelligent degree. In addition, flaw detection technology is also used to detect the integrity of the internal structure of the sucker rod, so as to timely discover and evaluate potential internal defects of the sucker rod, and provide a basis for subsequent repair work. In this way, comprehensive cleaning, detection and repair of the sucker rod are achieved, so as to improve the reliability and service life of the sucker rod. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0017] Figure 1 A schematic diagram of the application architecture of the sucker rod cleaning and repair method in an embodiment of the present application is shown.

[0018] Figure 2 A flow chart of a sucker rod cleaning and repairing method in an embodiment of the present application is shown.

[0019] Figure 3 A schematic diagram of the sucker rod process flow in an embodiment of the present application is shown.

[0020] Figure 4 A flowchart of sub-step S530 of the sucker rod cleaning and repairing method in an embodiment of the present application is shown.

[0021] Figure 5 A flow chart of sub-step S533 of the sucker rod cleaning and repairing method in an embodiment of the present application is shown.

[0022] Figure 6 A schematic structural diagram of a sucker rod cleaning and repair system in an embodiment of the present application is shown.

[0023] Figure 7 The application scenario diagram of the sucker rod cleaning and repair method in the embodiment of the present application is shown.

[0024] Figure 8 A schematic diagram of a storage medium according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0026] The terms used in this specification are those common terms currently widely used in the art in consideration of the functions of the present application, but these terms may vary according to the intention of a person of ordinary skill in the art, precedents, or new technologies in the art. In addition, specific terms may be selected by the applicant, and in this case, their detailed meanings will be described in the detailed description of the present application. Therefore, the terms used in the specification should not be understood as simple names, but rather as a general description based on the meaning of the terms and the present application.

[0027] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0028] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0029] Figure 1 A schematic diagram of the application architecture of the sucker rod cleaning and repair method in an embodiment of the present application is shown, including a server 100 and a terminal device 200.

[0030] The terminal device 200 and the server 100 can be connected via the Internet to achieve mutual communication. Optionally, the above-mentioned Internet uses standard communication technology and / or protocol. The Internet is usually the Internet, but it can also be any network, including but not limited to any combination of a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired or wireless network, a private network or a virtual private network. In some embodiments, the data exchanged through the network can be represented by technologies and / or formats including Hyper Text Markup Language (HTML), Extensible Markup Language (XML), etc. In addition, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), Internet Protocol Security (IPsec), etc. can be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can also be used to replace or supplement the above-mentioned data communication technologies.

[0031] The server 100 can provide various network services for the terminal device 200, wherein the server 100 can be a single server, a server cluster consisting of several servers, or a cloud computing center. Specifically, the server 100 may include a processor 110 (Center Processing Unit, CPU), a memory 120, an input device 130, and an output device 140, etc. The input device 130 may include a keyboard, a mouse, a touch screen, etc., and the output device 140 may include a display device, such as a liquid crystal display (Liquid Crystal Display, LCD), a cathode ray tube (Cathode Ray Tube, CRT), etc.

[0032] The memory 120 may include a read-only memory (ROM) and a random access memory (RAM), and provides the processor 110 with program instructions and data stored in the memory 120. In the embodiment of the present application, the memory 120 may be used to store the program of the sucker rod cleaning and repairing method in the embodiment of the present application.

[0033] The processor 110 calls the program instructions stored in the memory 120, and the processor 110 is used to execute the steps of any sucker rod cleaning and repair method in the embodiments of the present application according to the obtained program instructions.

[0034] In addition, the application architecture diagram in the embodiment of the present application is intended to more clearly illustrate the technical solution in the embodiment of the present application, and does not constitute a limitation on the technical solution provided in the embodiment of the present application. Of course, for other application architectures and business applications, the technical solution provided in the embodiment of the present application is also applicable to similar problems.

[0035] The following is a non-restrictive description of the sucker rod cleaning and repair method provided according to at least one embodiment of the present application through several examples or embodiments. As described below, different features in these specific examples or embodiments can be combined with each other without conflicting with each other to obtain new examples or embodiments, and these new examples or embodiments also fall within the scope of protection of the present application.

[0036] In view of the above technical problems, in the technical solution of this application, a sucker rod cleaning and repair method is proposed, such as Figure 2 As shown, it includes the following steps: S510, grabbing the sucker rod to be cleaned; S520, using a fully automatic closed high-temperature cleaning machine to remove rust, scale and wax from the sucker rod to be cleaned to obtain a cleaned sucker rod; S530, transferring the cleaned sucker rod to the flaw detection line to obtain a flaw detection result; S540, in response to the flaw detection result being unqualified, marking the cleaned sucker rod as unqualified; S550, in response to the flaw detection result being qualified, repairing the cleaned sucker rod to obtain a repaired sucker rod. In the above sucker rod cleaning and repair method, a fully automatic closed high-temperature cleaning machine is used, which can efficiently and automatically remove rust, scale and wax from the sucker rod to ensure the cleaning effect and intelligence. In addition, flaw detection technology is also used to detect the integrity of the internal structure of the sucker rod, so as to timely discover and evaluate potential internal defects of the sucker rod, and provide a basis for subsequent repair work. In this way, the sucker rod is fully cleaned, inspected and repaired, so as to improve the reliability and service life of the sucker rod. Figure 3 The whole process of sucker rod cleaning and repair is shown.

[0037] Among them, in step S510, the sucker rods are manually sorted, and the bent, deformed or damaged sucker rods are screened out and sent to the waste area. The qualified sucker rods screened out are placed on the rack to be cleaned by the pipe grabber. When the water temperature reaches above 85°C, they pass through the conveyor line into the fully automatic closed sucker rod hot cleaning machine for the cleaning process.

[0038] Among them, in step S520, a fully automatic closed sucker rod hot cleaning device adopts a central pipe groove drive, continuous feeding, synchronous discharging, and fully closed operation mode. The heating area of ​​the sucker rod is uniform, the operation is stable, and the heating time can be controlled according to the properties of the crude oil. The heating system of the cleaning machine uses a thermal energy generator to directly heat the cleaning medium (water). The heating speed is fast, the thermal efficiency utilization rate is high, and the temperature controllability is good. The cleaning process is automatically controlled to reduce the labor intensity of workers and prevent safety accidents. The internal structure of the cleaning equipment adopts closed sedimentation and automatic concentration for dirty oil and sludge. The sludge pool in the cleaning area is removed regularly or irregularly according to the production volume, and transported to the designated area in the factory for stacking, which greatly improves the control of sludge on environmental pollution and ensures the environmental protection requirements of the working environment.

[0039] It should be understood that magnetic flux leakage detection is a commonly used non-destructive testing technology that evaluates the integrity of the internal structure of the sucker rod by detecting the changes in the magnetic field on the surface of the sucker rod. Based on this, the technical concept of the present application is to collect the magnetic flux leakage detection signal of the cleaned sucker rod through magnetic flux leakage detection technology during the flaw detection of the cleaned sucker rod on the flaw detection line, and introduce a signal processing and analysis algorithm based on artificial intelligence and deep learning at the back end to analyze the magnetic flux leakage detection signal, so as to determine whether there are internal structural defects in the cleaned sucker rod. In this way, the intelligence level and reliability of sucker rod defect detection and flaw detection can be improved, providing more accurate data for the repair of sucker rods.

[0040] The HY-M-B30B series digital sucker rod nondestructive testing instrument can be used for testing. This instrument is a computerized nondestructive testing instrument for detecting surface cracks, corrosion pits and wear in sucker rods. It uses a local defect detection sensor. The detection signal is collected and judged by the computer after amplification, filtering and other processing, so as to realize the comprehensive detection of cracks, rust and rod wear. For the detection of cracks and rust, 8 channels are used for independent processing horizontally to achieve accurate classification of defects, reduce the misjudgment of useful sucker rods and the omission of scrapped sucker rods. The alarm threshold of the detection equipment is adjusted by software, and the signal status can be displayed in macro or detailed form to realize accurate flaw detection of the sucker rod state. The test results can be displayed, stored and printed. The sucker rod diameters used in this equipment are 19, 22 and 25 mm. The probe adopts a tile-shaped structure, and each specification corresponds to a probe. In the software, corresponding settings can be made according to quality requirements, such as adjusting the flaw detection sensitivity and alarm threshold, and classifying and recording the flaw detection results and reports. The sucker rods that pass the flaw detection will enter the manual wire inspection process, the sucker rods that fail the wire inspection will be scrapped, and the sucker rods that pass the wire inspection will enter the coupling replacement process.

[0041] Furthermore, if Figure 4As shown, according to the sucker rod cleaning and repair method of the embodiment of the present application, in step S530, the cleaned sucker rod is transferred to the flaw detection line to obtain the flaw detection result, including: S531, obtaining the leakage magnetic detection signal of the cleaned sucker rod; S532, performing local signal segment feature extraction on the leakage magnetic detection signal to obtain a sequence of current semantic association feature vectors of the leakage magnetic detection signal segments; S533, inputting the sequence of current semantic association feature vectors of the leakage magnetic detection signal segments into a self-supervised feature sequence significance fusion network to obtain a global leakage magnetic detection signal self-supervised significant fusion representation vector; S534, determining whether there is an internal structural defect based on the global leakage magnetic detection signal self-supervised significant fusion representation vector.

[0042] Specifically, in the process of transferring the cleaned sucker rod to the flaw detection line to obtain the flaw detection result, first, the leakage magnetic detection signal of the cleaned sucker rod is obtained. Next, considering that the leakage magnetic detection signal is obtained by a sensor, it contains the magnetic field information inside the sucker rod. These signals may be continuous and complex waveforms, containing a large amount of data information, and directly processing the entire signal may be very complicated. In order to better analyze and understand these signals, it is necessary to perform signal segmentation on them and decompose them into several fragments to simplify the processing process and enable independent feature extraction and analysis of each fragment. Based on this, in the technical solution of the present application, the leakage magnetic detection signal is segmented to obtain a sequence of leakage magnetic detection signal fragments. In particular, the signal segmentation processing method can also help to subsequently identify key areas in the signal, which may be related to specific defects of the sucker rod.

[0043] Then, the sequence of the magnetic flux leakage detection signal segments is subjected to feature mining in a signal feature extractor based on a one-dimensional convolutional neural network model to extract the implicit current semantic association feature information of each magnetic flux leakage detection signal segment in the sequence of the magnetic flux leakage detection signal segments, thereby obtaining a sequence of current semantic association feature vectors of the magnetic flux leakage detection signal segments. In magnetic flux leakage detection, the one-dimensional convolutional neural network model can automatically learn the local patterns and features in each magnetic flux leakage detection signal segment, which may be related to the defects of the sucker rod. In this way, it is helpful to highlight the feature information related to the sucker rod defect identification task by analyzing the correlation and difference between the local current semantics in each signal segment, and provide a more reliable basis for the subsequent flaw detection result analysis and judgment.

[0044] Correspondingly, in step S532, local signal segment feature extraction is performed on the leakage magnetic detection signal to obtain a sequence of current semantic association feature vectors of the leakage magnetic detection signal segments, including: signal segmentation of the leakage magnetic detection signal to obtain a sequence of leakage magnetic detection signal segments; passing the sequence of leakage magnetic detection signal segments through a signal feature extractor based on a one-dimensional convolutional neural network model to obtain a sequence of current semantic association feature vectors of the leakage magnetic detection signal segments.

[0045] Furthermore, considering that in the process of flaw detection of the sucker rod, the semantic features of each local signal segment in the leakage magnetic detection signal do not exist independently, these leakage magnetic detection signal segment current semantics have an overall correlation relationship and importance feature information. That is to say, each leakage magnetic detection signal segment current semantic association feature vector in the sequence of the leakage magnetic detection signal segment current semantic association feature vector constitutes the whole of the leakage magnetic detection signal semantic feature, and these signal local segment semantics have different importance for the subsequent flaw detection task in the fusion process. Based on this, in the technical solution of the present application, the sequence of the leakage magnetic detection signal segment current semantic association feature vector is further input into the self-supervised feature sequence significant fusion network to obtain the global leakage magnetic detection signal self-supervised significant fusion representation vector. Through the processing of the self-supervised feature sequence significant fusion network, it is possible to automatically learn and capture the mutual correlation relationship and importance semantics between the semantic features of each local signal segment in the leakage magnetic detection signal, and integrate these current semantic features from different signal segments to form a global and comprehensive feature representation. This integration method helps to capture the global patterns and anomalies in the leakage magnetic detection signal that may be ignored by a single segment. It is worth mentioning that the self-supervised feature sequence saliency fusion network can self-learn the correlation relationship and interaction characteristics between the current semantics of each local signal segment, so as to focus on identifying and emphasizing those features that have the greatest impact on the flaw detection results, namely, the saliency features. This helps to improve the accuracy of detection because the model can focus on the feature information that is most likely to indicate defects, providing a more accurate and comprehensive feature basis for the final defect judgment.

[0046] Accordingly, in step S533, if Figure 5As shown, the sequence of the magnetic flux leakage detection signal segment current semantic association feature vectors is input into the self-supervised feature sequence significance fusion network to obtain the global magnetic flux leakage detection signal self-supervised significant fusion representation vector, including: S5331, calculating the mean vector of the sequence of the magnetic flux leakage detection signal segment current semantic association feature vectors as the center of the magnetic flux leakage detection signal segment current semantic distribution cluster to obtain the center feature vector of the magnetic flux leakage detection signal segment current semantic distribution cluster; S5332, calculating the relationship between each magnetic flux leakage detection signal segment current semantic association feature vector in the sequence of the magnetic flux leakage detection signal segment current semantic association feature vector and the center of the magnetic flux leakage detection signal segment current semantic distribution cluster. S5333, normalizing the sequence of similarity score values ​​of the centers of the current semantic distribution clusters of the leakage magnetic detection signal segments to obtain a sequence of similarity attention weight coefficients of the centers of the current semantic distribution clusters of the leakage magnetic detection signal segments; S5334, using the sequence of similarity attention weight coefficients of the centers of the current semantic distribution clusters of the leakage magnetic detection signal segments as weights, calculating the vector-by-vector weighted sum of the sequence of current semantically associated feature vectors of the leakage magnetic detection signal segments to obtain the self-supervised significant fusion representation vector of the global leakage magnetic detection signal.

[0047] Wherein, in step S5332, the semantic similarity score values ​​between each leakage magnetic detection signal segment current semantic association feature vector in the sequence of the leakage magnetic detection signal segment current semantic association feature vector and the central feature vector of the leakage magnetic detection signal segment current semantic distribution cluster are calculated to obtain a sequence of central similarity score values ​​of the leakage magnetic detection signal segment current semantic distribution cluster, including: respectively calculating the matrix product between each weight coefficient matrix and the corresponding leakage magnetic detection signal segment current semantic association feature vector to obtain a sequence of leakage magnetic detection signal segment current semantic weighted feature vectors; adding the corresponding bias vector to each leakage magnetic detection signal segment current semantic weighted feature vector in the sequence of the leakage magnetic detection signal segment current semantic weighted feature vector to obtain a sequence of leakage magnetic detection signal segment current semantic weighted bias adjusted feature vectors; inputting the sequence of leakage magnetic detection signal segment current semantic weighted bias adjusted feature vectors The function performs activation processing to obtain a sequence of magnetic flux leakage detection signal segment current semantic activation feature vectors; the product of each magnetic flux leakage detection signal segment current semantic activation feature vector in the sequence of magnetic flux leakage detection signal segment current semantic activation feature vectors and the central feature vector of the magnetic flux leakage detection signal segment current semantic distribution cluster is calculated respectively to obtain a sequence of similarity score values ​​of the central similarity of the magnetic flux leakage detection signal segment current semantic distribution cluster.

[0048] Among them, in step S5333, the sequence of similarity score values ​​of the current semantic distribution cluster centers of the leakage magnetic detection signal segments is normalized to obtain a sequence of similarity attention weight coefficients of the current semantic distribution cluster centers of the leakage magnetic detection signal segments, including: masking the respective similarity score values ​​of the current semantic distribution cluster centers of the leakage magnetic detection signal segments in the sequence of similarity score values ​​of the current semantic distribution cluster centers to obtain a sequence of masked similarity score values ​​of the current semantic distribution cluster centers of the leakage magnetic detection signal segments; masking the respective masked similarity score values ​​of the current semantic distribution cluster centers of the masked leakage magnetic detection signal segments in the sequence of similarity score values ​​of the current semantic distribution cluster centers by A soft maximum normalization process is performed in the function to obtain a sequence of similarity attention weight coefficients of the center of the current semantic distribution cluster of the leakage magnetic detection signal segment.

[0049] Among them, each leakage magnetic detection signal segment current semantic distribution cluster center similarity score value in the sequence of leakage magnetic detection signal segment current semantic distribution cluster center similarity score values ​​is masked to obtain a sequence of masked leakage magnetic detection signal segment current semantic distribution cluster center similarity score values, including: in response to each leakage magnetic detection signal segment current semantic distribution cluster center similarity score value in the sequence of leakage magnetic detection signal segment current semantic distribution cluster center similarity score value being greater than or equal to a predetermined threshold, each leakage magnetic detection signal segment current semantic distribution cluster center similarity score value in the sequence of leakage magnetic detection signal segment current semantic distribution cluster center similarity score value is taken as the original value, otherwise it is set to zero.

[0050] In a specific example, the sequence of the semantically associated feature vectors of the magnetic flux leakage detection signal segments is input into a self-supervised feature sequence significance fusion network to obtain a global magnetic flux leakage detection signal self-supervised significant fusion representation vector, including: inputting the sequence of the semantically associated feature vectors of the magnetic flux leakage detection signal segments into the self-supervised feature sequence significance fusion network and processing them with the following self-supervised feature significance fusion formula to obtain the global magnetic flux leakage detection signal self-supervised significant fusion representation vector; wherein the self-supervised feature significance fusion formula is: ;in, is the first in the sequence of the current semantic association feature vectors of the magnetic flux leakage detection signal segment The current semantic association feature vector of the leakage magnetic detection signal segment, is the number of vectors in the sequence of the current semantic association feature vectors of the magnetic flux leakage detection signal segment, is the central feature vector of the current semantic distribution cluster of the leakage magnetic detection signal segment, represents the transpose of a vector, represents the weight coefficient matrix, is the bias vector, express function, For the said The semantic similarity score between the semantic association feature vector of the magnetic flux leakage detection signal segment current and the central feature vector of the magnetic flux leakage detection signal segment current semantic distribution cluster, For masking operation, For the said The masked semantic similarity score of the current semantic association feature vector of the leakage magnetic detection signal segment, express function, It is the self-supervised significant fusion representation vector of the global magnetic flux leakage detection signal.

[0051] Then, the global magnetic flux leakage detection signal self-supervised significant fusion representation vector is passed through the classifier-based flaw detection result generator to obtain the flaw detection result, which is used to indicate whether there are internal structural defects. In other words, the global semantic significant fusion features in the magnetic flux leakage detection signal are used for classification processing to determine whether there are internal structural defects in the sucker rod after cleaning. In this way, the intelligence level and reliability of sucker rod defect detection and flaw detection can be improved, providing more accurate data for the repair of sucker rods.

[0052] Accordingly, in step S534, based on the self-supervised significant fusion representation vector of the global magnetic flux leakage detection signal, it is determined whether there is an internal structural defect, including: passing the self-supervised significant fusion representation vector of the global magnetic flux leakage detection signal through a classifier-based flaw detection result generator to obtain a flaw detection result, and the flaw detection result is used to indicate whether there is an internal structural defect.

[0053] It should be understood that the role of the classifier is to use the given categories and known training data to learn classification rules and classifiers, and then classify (or predict) unknown data. Logistic regression, SVM, etc. are often used to solve binary classification problems. For multi-class classification problems, logistic regression or SVM can also be used, but multiple binary classifications are required to form a multi-classification, but this is prone to errors and inefficient. Commonly used multi-classification methods include the Softmax classification function.

[0054] In a preferred embodiment, the sequence of the current semantic association feature vectors of the leakage magnetic detection signal segments expresses the current semantic association features of the leakage magnetic detection signal in the local time domain. When the sequence is input into the self-supervised feature sequence significance fusion network for feature fusion in the global time domain, due to the non-correspondence of the time series association feature distributions of each local time domain, the global leakage magnetic detection signal self-supervised significant fusion representation vector will have a global time domain time series feature distribution knowledge offset based on the significant difference in the local time domain time series sequence distribution, thereby causing the global time domain time series feature distribution to lack the class regression interpretability of the classifier, affecting the accuracy of the classification result.

[0055] Based on this, the present application optimizes the global magnetic flux leakage detection signal self-supervised significant fusion representation vector when passing the global magnetic flux leakage detection signal self-supervised significant fusion representation vector through a classifier-based flaw detection result generator to obtain a flaw detection result, including the steps of: calculating the autocorrelation matrix of the global magnetic flux leakage detection signal self-supervised significant fusion representation vector, that is: ;in, represents the global magnetic flux leakage detection signal self-supervised significant fusion representation vector, represents the transpose of a vector, represents matrix multiplication, Denotes the autocorrelation matrix.

[0056] The global magnetic flux leakage detection signal self-supervised significant fusion representation vector is query-multiplied with the self-correlation matrix to obtain an intermediate feature vector, that is: ;in, represents the global magnetic flux leakage detection signal self-supervised significant fusion representation vector, represents the autocorrelation matrix, represents matrix multiplication, Denotes the intermediate eigenvector.

[0057] Calculate the autocorrelation matrix Row vector and Inner product sum of row vectors distance, to obtain the first weight matrix and the second weight matrix , expressed as: ;in, and The autocorrelation matrix is ​​represented by Row vector and row vector, represents the transpose of a vector, represents matrix multiplication, represents the autocorrelation matrix, represents the first weight matrix, represents the first weight matrix The eigenvalues ​​of the positions, represents the second weight matrix, represents the second weight matrix The eigenvalues ​​of the positions, Represents the calculation vector distance.

[0058] The first weight matrix and the second weight matrix After matrix multiplication, further matrix multiplication is performed with the intermediate feature vector to obtain the optimized global magnetic flux leakage detection signal self-supervised significant fusion representation vector, that is: ;in, represents the first weight matrix, represents the second weight matrix, represents the intermediate eigenvector, represents matrix multiplication, The optimized global magnetic flux leakage detection signal self-supervised significant fusion representation vector is represented. Finally, the optimized global magnetic flux leakage detection signal self-supervised significant fusion representation vector is passed through a flaw detection result generator based on a classifier to obtain a flaw detection result.

[0059] That is, for the bit-by-bit scattering correlation response space of the eigenvalues ​​of the self-supervised significant fusion representation vector of the global magnetic flux leakage detection signal, a deep embedding structure in the high-order correlation space is obtained through the joint embedding space of the vector inner product embedding space and the vector independent embedding space, and the decomposable correlation response sub-dimensions of the bit-by-bit scattering correlation response space of the eigenvalues ​​of the self-supervised significant fusion representation vector of the global magnetic flux leakage detection signal are represented as a set of spatial sub-dimensions to further perform dependent spatial mapping of the decomposable correlation response sub-dimensions of the deep embedding structure of the self-supervised significant fusion representation vector of the global magnetic flux leakage detection signal, so as to improve the interpretability of the solution space of the class regression solution of the self-supervised significant fusion representation vector of the global magnetic flux leakage detection signal based on the correlation distribution through the complex dimensional representation of the correlation response, thereby improving the accuracy of the flaw detection result obtained by the self-supervised significant fusion representation vector of the global magnetic flux leakage detection signal through the classifier-based flaw detection result generator.

[0060] Furthermore, in the technical solution of the present application, the sucker rod cleaning and repair method also includes a training step: for training the signal feature extractor based on the one-dimensional convolutional neural network model, the self-supervised feature sequence significance fusion network and the classifier-based flaw detection result generator.

[0061] The training step includes: obtaining training data, wherein the training data includes the training magnetic flux leakage detection signal of the sucker rod after cleaning and the actual flaw detection result; performing signal segmentation on the training magnetic flux leakage detection signal to obtain a sequence of training magnetic flux leakage detection signal segments; passing the sequence of training magnetic flux leakage detection signal segments through the signal feature extractor based on the one-dimensional convolutional neural network model to obtain a sequence of current semantic association feature vectors of the training magnetic flux leakage detection signal segments; inputting the sequence of current semantic association feature vectors of the training magnetic flux leakage detection signal segments into the self-supervised feature sequence significance fusion network to obtain a training global magnetic flux leakage detection signal self-supervised significant fusion representation vector ; Optimize the self-supervised significant fusion representation vector of the training global magnetic flux leakage detection signal to obtain an optimized self-supervised significant fusion representation vector of the training global magnetic flux leakage detection signal; Pass the optimized self-supervised significant fusion representation vector of the training global magnetic flux leakage detection signal through the classifier-based flaw detection result generator to obtain a training flaw detection result; Calculate the cross entropy loss function value between the training flaw detection result and the true flaw detection result to obtain a classification loss function value; Based on the classification loss function value, train the signal feature extractor based on the one-dimensional convolutional neural network model, the self-supervised feature sequence significance fusion network and the classifier-based flaw detection result generator.

[0062] In a preferred embodiment, when the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector is passed through a classifier-based flaw detection result generator to obtain a training flaw detection result, the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector is optimized. Specifically, the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector is optimized to obtain an optimized training global magnetic flux leakage detection signal self-supervised significant fusion representation vector, including the steps of: determining the defect class probability value corresponding to the presence of internal structural defects and the non-defect class probability value corresponding to the absence of internal structural defects obtained by inputting the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector into the classifier-based flaw detection result generator; respectively calculating the power function of each eigenvalue of the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector with the defect class probability value as an exponent and the power function with the non-defect class probability value as an exponent to obtain the training global magnetic flux leakage detection signal self-supervised significant fusion defect class vector and the training global magnetic flux leakage detection signal self-supervised significant fusion non-defect class vector; The defect class probability value is point-multiplied with the training global magnetic flux leakage detection signal self-supervised significant fusion non-defect class vector to obtain a first training global magnetic flux leakage detection signal self-supervised significant fusion cross-class vector, and the non-defect class probability value is point-multiplied with the training global magnetic flux leakage detection signal self-supervised significant fusion defect class vector to obtain a second training global magnetic flux leakage detection signal self-supervised significant fusion cross-class vector; and, after point-multiplying the first training global magnetic flux leakage detection signal self-supervised significant fusion cross-class vector with the second training global magnetic flux leakage detection signal self-supervised significant fusion cross-class vector, further point-multiplying the defect class probability value with the training global magnetic flux leakage detection signal self-supervised significant fusion defect class vector to obtain an optimized training global magnetic flux leakage detection signal self-supervised significant fusion representation vector.

[0063] Through the above optimization steps, when the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector is classified through the classifier-based flaw detection result generator, in order to achieve unsupervised domain adaptation from the feature distribution domain of the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector to the probability distribution domain of the classification probability, the defect class probability value and the non-defect class probability value obtained by the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector through the classifier-based flaw detection result generator are used as domain agents, and the moving average of the probability distribution of the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector is performed through the interaction of the power function class distribution as an exponent, and the knowledge transfer from the unlabeled classification feature to the labeled probability distribution is achieved by superimposing the feature domain probability distribution knowledge of the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector itself, thereby promoting the classification operation of the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector through the classifier-based flaw detection result generator, that is, improving the classification training speed of the classifier-based flaw detection result generator and the accuracy of the classification results. In this way, it is possible to further accurately determine whether the sucker rod has internal structural defects after cleaning, providing more accurate data for the repair work of the sucker rod.

[0064] Based on the above embodiments, see Figure 6 The figure is a schematic diagram of the structure of a sucker rod cleaning and repairing system 800 in the embodiment of the present application. The sucker rod cleaning and repairing system 800 includes: a sucker rod grabbing module 810 for grabbing the sucker rod to be cleaned; a cleaning module 820 for removing rust, scale and wax from the sucker rod to be cleaned by using a fully automatic closed high-temperature cleaning machine to obtain a cleaned sucker rod; a flaw detection module 830 for transferring the cleaned sucker rod to a flaw detection line to obtain a flaw detection result; an identification module 840 for identifying the cleaned sucker rod as unqualified in response to the flaw detection result being unqualified; and a repair module 850 for repairing the cleaned sucker rod to obtain a repaired sucker rod in response to the flaw detection result being qualified.

[0065] Here, those skilled in the art will appreciate that the specific functions and operations of each module in the above-mentioned sucker rod cleaning and repair system 800 have been described in detail above. Figures 2 to 5 The sucker rod cleaning and repair method has been described in detail, and therefore, its repeated description will be omitted.

[0066] Figure 7 FIG. 1 is an application scenario diagram of the sucker rod cleaning and repair method according to an embodiment of the present application. Figure 7 As shown, in this application scenario, first, the magnetic flux leakage detection signal of the sucker rod after cleaning is obtained (for example, Figure 7D), then, the magnetic flux leakage detection signal is input to a server (for example, Figure 7 In S) as shown in , the server can use the sucker rod cleaning and repair algorithm to process the magnetic leakage detection signal to obtain a flaw detection result indicating whether there is an internal structural defect.

[0067] Based on the above embodiment, another exemplary embodiment of an electronic device is provided in the embodiment of the present application. In some possible implementations, the electronic device in the embodiment of the present application may include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor can implement the steps of the sucker rod cleaning and repair method in the above embodiment when executing the program.

[0068] For example, in the case of electronic equipment Figure 1 Taking the server 100 in the example as an example, the processor in the electronic device is the processor 110 in the server 100, and the memory in the electronic device is the memory 120 in the server 100.

[0069] An embodiment of the present application also provides a computer-readable storage medium. Figure 8 1 shows a schematic diagram of a computer-readable storage medium 1000 according to an embodiment of the present application. Figure 8 As shown, the computer-readable storage medium 1000 stores computer-executable instructions 1001. When the computer-executable instructions 1001 are executed by the processor, the sucker rod cleaning and repair method according to the embodiment of the present application described with reference to the above figures can be executed. The computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc.

[0070] The embodiment of the present application also provides a computer program product or a computer program, which includes computer executable instructions, and the computer executable instructions are stored in a computer readable storage medium. The processor of the computer device reads the computer executable instructions from the computer readable storage medium, and the processor executes the computer executable instructions, so that the computer device executes the sucker rod cleaning and repair method according to the embodiment of the present application.

[0071] Those skilled in the art will appreciate that the contents disclosed in this application may be subject to various variations and improvements. For example, the various devices or components described above may be implemented by hardware, or by software, firmware, or a combination of some or all of the three.

[0072] In addition, although the present application makes various references to certain units in the system according to embodiments of the present application, any number of different units can be used and run on the client and / or server. The units are only illustrative, and different aspects of the system and method can use different units.

[0073] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software function modules. The present application is not limited to any particular form of combination of hardware and software.

[0074] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0075] The above is an explanation of the present application and should not be considered as a limitation thereof. Although several exemplary embodiments of the present application have been described, it will be readily understood by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application. Therefore, all of these modifications are intended to be included within the scope of the present application as defined by the claims. It should be understood that the above is an explanation of the present application and should not be considered to be limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present application is defined by the claims and their equivalents.

Claims

1. A method for cleaning and repairing a sucker rod, characterized in that: include: Grab the sucker rod to be cleaned; Using a fully automatic closed high-temperature cleaning machine to remove rust, scale and wax from the sucker rod to be cleaned to obtain a cleaned sucker rod; Transferring the cleaned sucker rod to a flaw detection line to obtain a flaw detection result; In response to the flaw detection result being unqualified, marking the cleaned sucker rod as unqualified; In response to the flaw detection result being qualified, repairing the cleaned sucker rod to obtain a repaired sucker rod; The cleaned sucker rod is transferred to a flaw detection line to obtain a flaw detection result, including: Acquiring a magnetic flux leakage detection signal of the sucker rod after cleaning; Extracting local signal segment features of the magnetic flux leakage detection signal to obtain a sequence of current semantic association feature vectors of the magnetic flux leakage detection signal segments; Inputting the sequence of the current semantic association feature vectors of the leakage magnetic detection signal segments into a self-supervised feature sequence significance fusion network to obtain a global leakage magnetic detection signal self-supervised significant fusion representation vector; Determine whether there is an internal structural defect based on the self-supervised significant fusion representation vector of the global magnetic flux leakage detection signal; The sequence of the current semantic association feature vectors of the magnetic flux leakage detection signal segments is input into a self-supervised feature sequence saliency fusion network to obtain a global magnetic flux leakage detection signal self-supervised saliency fusion representation vector, including: Calculating the mean vector of the sequence of the magnetic flux leakage detection signal segment current semantic association feature vectors as the magnetic flux leakage detection signal segment current semantic distribution cluster center to obtain the magnetic flux leakage detection signal segment current semantic distribution cluster center feature vector; Calculating the semantic similarity score between each magnetic flux leakage detection signal segment current semantic association feature vector in the sequence of magnetic flux leakage detection signal segment current semantic association feature vectors and the central feature vector of the magnetic flux leakage detection signal segment current semantic distribution cluster to obtain a sequence of central similarity score values ​​of the magnetic flux leakage detection signal segment current semantic distribution cluster; Normalizing the sequence of similarity score values ​​of the current semantic distribution cluster centers of the magnetic flux leakage detection signal segments to obtain a sequence of similarity attention weight coefficients of the current semantic distribution cluster centers of the magnetic flux leakage detection signal segments; Taking the sequence of similarity attention weight coefficients of the center of the current semantic distribution cluster of the leakage magnetic detection signal segment as the weight, the vector-by-vector weighted sum of the sequence of current semantic association feature vectors of the leakage magnetic detection signal segment is calculated to obtain the global leakage magnetic detection signal self-supervised significant fusion representation vector.

2. The method for cleaning and repairing a sucker rod according to claim 1, characterized in that: Extracting local signal segment features of the magnetic flux leakage detection signal to obtain a sequence of current semantic association feature vectors of the magnetic flux leakage detection signal segments includes: Performing signal segmentation on the magnetic flux leakage detection signal to obtain a sequence of magnetic flux leakage detection signal segments; The sequence of the magnetic flux leakage detection signal segments is passed through a signal feature extractor based on a one-dimensional convolutional neural network model to obtain a sequence of current semantic association feature vectors of the magnetic flux leakage detection signal segments.

3. The method for cleaning and repairing a sucker rod according to claim 2, characterized in that: Calculating the semantic similarity score between each magnetic flux leakage detection signal segment current semantic association feature vector in the sequence of magnetic flux leakage detection signal segment current semantic association feature vectors and the magnetic flux leakage detection signal segment current semantic distribution cluster center feature vector to obtain a sequence of magnetic flux leakage detection signal segment current semantic distribution cluster center similarity score values, including: Calculating the matrix products of the weight coefficient matrices and the corresponding magnetic flux leakage detection signal segment current semantics associated feature vectors to obtain a sequence of magnetic flux leakage detection signal segment current semantics weighted feature vectors; Adding a corresponding bias vector to each magnetic flux leakage detection signal segment current semantic weighted feature vector in the sequence of magnetic flux leakage detection signal segment current semantic weighted feature vectors to obtain a sequence of magnetic flux leakage detection signal segment current semantic weighted bias adjusted feature vectors; The sequence of the current semantic weighted bias adjusted feature vectors of the magnetic flux leakage detection signal segments is input The function performs activation processing to obtain a sequence of semantic activation feature vectors of the current of the leakage magnetic detection signal segment; The products of each magnetic flux leakage detection signal segment current semantic activation feature vector in the sequence of magnetic flux leakage detection signal segment current semantic activation feature vectors and the central feature vector of the magnetic flux leakage detection signal segment current semantic distribution cluster are calculated respectively to obtain a sequence of similarity score values ​​of the central current semantic distribution cluster of the magnetic flux leakage detection signal segment current.

4. The method for cleaning and repairing a sucker rod according to claim 3, characterized in that: The sequence of similarity score values ​​of the current semantic distribution cluster centers of the magnetic flux leakage detection signal segments is normalized to obtain a sequence of similarity attention weight coefficients of the current semantic distribution cluster centers of the magnetic flux leakage detection signal segments, including: Masking each magnetic flux leakage detection signal segment current semantic distribution cluster center similarity score value in the sequence of magnetic flux leakage detection signal segment current semantic distribution cluster center similarity score values ​​to obtain a sequence of masked magnetic flux leakage detection signal segment current semantic distribution cluster center similarity score values; The similarity scores of the centers of the current semantic distribution clusters of the masked magnetic flux leakage detection signal segments in the sequence of the similarity scores of the centers of the current semantic distribution clusters of the masked magnetic flux leakage detection signal segments are calculated by A soft maximum normalization process is performed in the function to obtain a sequence of similarity attention weight coefficients of the center of the current semantic distribution cluster of the leakage magnetic detection signal segment.

5. The method for cleaning and repairing a sucker rod according to claim 4, characterized in that: Masking each magnetic flux leakage detection signal segment current semantic distribution cluster center similarity score value in the sequence of magnetic flux leakage detection signal segment current semantic distribution cluster center similarity score values ​​to obtain a sequence of masked magnetic flux leakage detection signal segment current semantic distribution cluster center similarity score values, including: In response to the fact that each similarity score value of the center of the current semantic distribution cluster of the leakage magnetic detection signal segment in the sequence of similarity score values ​​of the current semantic distribution cluster of the leakage magnetic detection signal segment is greater than or equal to a predetermined threshold, each similarity score value of the center of the current semantic distribution cluster of the leakage magnetic detection signal segment in the sequence of similarity score values ​​of the current semantic distribution cluster of the leakage magnetic detection signal segment is taken as the original value, otherwise it is set to zero.

6. The method for cleaning and repairing a sucker rod according to claim 5, characterized in that: Based on the global magnetic flux leakage detection signal self-supervised significant fusion representation vector, determining whether there is an internal structural defect, including: The global magnetic flux leakage detection signal self-supervised significant fusion representation vector is passed through a classifier-based flaw detection result generator to obtain a flaw detection result, and the flaw detection result is used to indicate whether there is an internal structural defect.

7. The method for cleaning and repairing a sucker rod according to claim 6, characterized in that: The method further includes a training step for training the signal feature extractor based on the one-dimensional convolutional neural network model, the self-supervised feature sequence significance fusion network and the flaw detection result generator based on the classifier; Wherein, the training step includes: Acquiring training data, wherein the training data includes a training magnetic leakage detection signal of the sucker rod after cleaning and a real flaw detection result; Performing signal segmentation on the training magnetic flux leakage detection signal to obtain a sequence of training magnetic flux leakage detection signal segments; Passing the sequence of training magnetic flux leakage detection signal segments through the signal feature extractor based on the one-dimensional convolutional neural network model to obtain a sequence of current semantic association feature vectors of the training magnetic flux leakage detection signal segments; Inputting the sequence of the training magnetic flux leakage detection signal segment current semantic association feature vectors into the self-supervised feature sequence significance fusion network to obtain the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector; Optimizing the training global magnetic flux leakage detection signal self-supervised significant fusion representation vector to obtain an optimized training global magnetic flux leakage detection signal self-supervised significant fusion representation vector; Passing the optimized training global magnetic flux leakage detection signal self-supervised significant fusion representation vector through the classifier-based flaw detection result generator to obtain a training flaw detection result; Calculating a cross entropy loss function value between the training flaw detection result and the actual flaw detection result to obtain a classification loss function value; The signal feature extractor based on the one-dimensional convolutional neural network model, the self-supervised feature sequence significance fusion network and the classifier-based flaw detection result generator are trained based on the classification loss function value.

Images