Pipeline defect classification and identification method

The pre-trained model is used to identify and classify oil and gas pipeline defects, which solves the problem of poor accuracy of magnetic leakage detection, and realizes the accurate identification and classification of common pipeline defects, improving the accuracy and efficiency of detection.

CN118968494AInactive Publication Date: 2024-11-15PIPECHINA SOUTH CHINA CO +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202410956440.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-17
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing magnetic leakage detection methods have poor accuracy in pipeline defect identification, and a single-stage object detection algorithm may damage the accuracy and reliability of magnetic leakage detection.

Method used

The pre-trained model is used to identify and classify oil and gas pipeline defects, and the classification recognition model is trained through feature extractors and classifiers to improve the accuracy and efficiency of leakage detection, so as to achieve common pipeline defects (such as pitting, grooves, and grooves). , accurate classification of general corrosion).

Benefits of technology

It significantly improves the average accuracy and recall rate of leakage magnetic detection, achieving 30% and 60% improvements respectively, with faster running speeds and only a few seconds for defect recognition of a single image, which is better than other popular image classification algorithms.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968494A_ABST
    Figure CN118968494A_ABST
Patent Text Reader

Abstract

The embodiment of the invention provides a pipeline defect classification and identification method, which can solve the technical problem of poor accuracy of existing magnetic flux leakage detection. The pipeline defect classification and identification method comprises the following steps: acquiring defect data of a to-be-trained pipeline; based on the to-be-trained pipeline defect data, training the pre-training model to obtain a feature extractor; and training a classification recognition model based on the training data set according to the feature extractor and the classifier. The invention relates to a method for accurately identifying and classifying internal defects of an oil and gas pipeline by using a pre-training model, which is used for accurately classifying common pipeline defects (such as pitting corrosion, grooves, grooves, general corrosion and the like) so as to avoid pipeline fracture and leakage possibly caused by the defects. The pre-training model recognizes the type of the defect, the average precision and the recall rate are remarkably higher, and the improvement of 30% and 60% is achieved respectively. Therefore, the accuracy of magnetic flux leakage detection is improved. And moreover, the operation speed is relatively high, and the defect identification of a single image only needs several seconds.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of oil and gas pipelines, and in particular to a pipeline defect classification and identification method. Background Art

[0002] The complex environment of long-distance pipelines makes them prone to surface defects, posing a major threat to pipeline safety. Magnetic Flux Leakage Testing (MFL) is a non-destructive testing method that can be used to map and monitor pipeline defects.

[0003] Currently, MFL detects defects by applying a magnetic field to the surface of the object being inspected and then observing the leakage of the magnetic field. When the magnetic field encounters the defect, a change in magnetic resistance occurs, resulting in magnetic leakage. Further, based on feature extraction and data analysis, the location, shape and size of the defect can be determined by detecting this magnetic leakage.

[0004] However, the data processing and analysis methods used based on single-stage target detection algorithms may seriously damage the accuracy and reliability of magnetic leakage detection. Summary of the invention

[0005] The embodiment of the present application provides a pipeline defect classification and identification method, which can solve the technical problem of poor accuracy of existing magnetic leakage detection.

[0006] To achieve the above objectives, the embodiments of the present application adopt the following technical solutions:

[0007] In a first aspect, an embodiment of the present application provides a pipeline defect classification and identification method, which includes: obtaining pipeline defect data to be trained; training a pre-trained model based on the pipeline defect data to be trained to obtain a feature extractor; and training a classification and identification model based on a training data set according to the feature extractor and the classifier.

[0008] Based on the above description of the pipeline defect classification and identification method provided in the embodiment of the present application, it can be known that the pipeline defect classification and identification method includes a method for accurately identifying and classifying internal defects of oil and gas pipelines using a pre-trained model, which is used to accurately classify common pipeline defects (such as pitting, grooves, grooves, general corrosion, etc.), thereby avoiding pipeline ruptures and leakages that may be caused by these defects. The pre-trained model identifies the type of defects and has significantly higher average precision and recall, achieving an improvement of 30% and 60% respectively. In this way, the accuracy of magnetic flux leakage detection is improved.

[0009] As well, the pre-trained model also outperforms other popular image classification algorithms. It runs relatively fast, with defect identification taking only a few seconds for a single image.

[0010] In a feasible implementation manner of the first aspect, the pipeline defect data to be trained includes a training set and a test set, and the training set and the test set both include at least part of the pipeline defect data to be trained and a plurality of images without pipeline defects.

[0011] In a feasible implementation of the first aspect, when executing the step of training a pre-trained model based on the pipeline defect data to be trained to obtain a feature extractor, the pipeline defect classification and identification method also includes: inputting the pipeline defect data to be trained into a block linear projection module to obtain multiple data blocks; preprocessing the data blocks based on image segmentation, position embedding and category embedding to obtain preprocessed data blocks; and inputting the preprocessed data blocks into a transformer encoder.

[0012] In a feasible implementation of the first aspect, when executing the step of inputting the preprocessed data block into the transformer encoder, the pipeline defect classification and identification method also includes: training the preprocessed data block based on embedded image blocks, normalization, multi-head attention, normalization and multi-layer perceptron in sequence.

[0013] In a feasible implementation of the first aspect, when executing the steps of training the preprocessed data block based on the embedded image block, normalization, multi-head attention, normalization and multi-layer perceptron in sequence, the pipeline defect classification and identification method also includes: inputting the preprocessed data block into the embedded image block to obtain a first data result; normalizing the first data result in sequence and performing multi-head attention to obtain a second data result; combining the first data result and the second data result to obtain a third data result; normalizing the third data result and inputting it into the multi-layer perceptron to obtain a fourth data result; combining the third data result and the fourth data result to obtain a fifth data result.

[0014] In a feasible implementation of the first aspect, when executing the step of acquiring pipeline defect data to be trained, the pipeline defect classification and identification method further includes: acquiring pipeline defect data to be trained based on two pipeline pull-through experiments.

[0015] In a feasible implementation manner of the first aspect, the intersection-over-union ratio threshold is set to 0.5.

[0016] In a feasible implementation of the first aspect, the pipeline defect classification and identification method further includes: constructing a confusion matrix to evaluate the classification accuracy of the classification and identification model.

[0017] In a second aspect, an embodiment of the present application provides a pipeline defect classification and identification system, which includes: at least one processor; a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method provided in the first aspect.

[0018] The pipeline defect classification and identification system uses a pre-trained model to accurately identify and classify internal defects of oil and gas pipelines by executing the method provided in the first aspect, and is used to accurately classify common pipeline defects (such as pitting, grooves, grooves, general corrosion, etc.), thereby avoiding pipeline ruptures and leakages that may be caused by these defects. The pre-trained model identifies the type of defects and has significantly higher average precision and recall, achieving an improvement of 30% and 60% respectively. In this way, the accuracy of magnetic flux leakage detection is improved. In addition, the pre-trained model is also superior to other popular image classification algorithms. It runs relatively fast, and defect recognition for a single image only takes a few seconds.

[0019] In a third aspect, an embodiment of the present application provides a computer-readable medium having computer program instructions stored thereon, and the computer program instructions can be executed by a processor to implement the method provided in the first aspect.

[0020] The computer program instructions in the computer-readable medium implement the method provided in the first aspect, and use the pre-trained model to accurately identify and classify the internal defects of the oil and gas pipeline, so as to accurately classify common pipeline defects (such as pitting, grooves, grooves, general corrosion, etc.), thereby avoiding pipeline rupture and leakage that may be caused by these defects. The pre-trained model identifies the type of defects and has significantly higher average precision and recall, achieving an improvement of 30% and 60% respectively. In this way, the accuracy of magnetic flux leakage detection is improved. In addition, the pre-trained model is also superior to other popular image classification algorithms. It runs relatively fast, and defect recognition of a single image only takes a few seconds. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of the structure of a pipeline defect classification and identification system provided in an embodiment of the present application;

[0022] Figure 2 A schematic diagram of a pipeline defect classification and identification method provided in an embodiment of the present application;

[0023] Figure 3 A schematic diagram of the structure of a robot that obtains pipeline defect data to be trained in a pipeline defect classification and recognition method provided in an embodiment of the present application;

[0024] Figure 4 A schematic diagram of the structure of a pre-trained model in a pipeline defect classification and identification method provided in an embodiment of the present application;

[0025] Figure 5 Four example images of defect images input into a pre-trained model in a pipeline defect classification and recognition method provided in an embodiment of the present application;

[0026] Figure 6A loss or accuracy curve diagram of a training and verification process in a pipeline defect classification and identification method provided in an embodiment of the present application;

[0027] Figure 7 A schematic diagram of a confusion matrix diagram in a pipeline defect classification and identification method provided in an embodiment of the present application;

[0028] Figure 8 A graph showing the accuracy of different types of defect recognition calculated in a pipeline defect classification and recognition method provided in an embodiment of the present application;

[0029] Fig. 9 Performance graphs for different deep learning architectures;

[0030] Fig.10 A comparison chart of different deep learning architectures. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present invention will be described below in conjunction with the accompanying drawings in the embodiments of the present invention. In the description of the embodiments of the present invention, unless otherwise specified, "multiple" refers to two or more than two. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0032] In addition, in order to facilitate the clear description of the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, the words "first", "second" and the like are used to distinguish the same items or similar items with substantially the same functions and effects. Those skilled in the art will understand that the words "first", "second" and the like do not limit the quantity and execution order, and the words "first", "second" and the like do not necessarily limit the difference. At the same time, in the embodiments of the present invention, the words "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or design. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a concrete way for easy understanding.

[0033] The principles and features of the present application are described below. The examples given are only used to explain the present application and are not used to limit the scope of the present application.

[0034] The embodiment of the present application provides a pipeline defect classification and identification method for accurately classifying common pipeline defects (such as pitting, grooves, grooves, general corrosion, etc.) to avoid pipeline rupture and leakage that may be caused by these defects.

[0035] The embodiment of the present application provides a pipeline defect classification and identification system, which can execute the pipeline defect classification and identification method provided in the embodiment of the present application. Figure 1 A schematic diagram of the structure of a pipeline defect classification and identification system provided in an embodiment of the present application.

[0036] like Figure 1 As shown, the pipeline defect classification and identification system 001 includes at least one processor 011 and a memory 012 that is communicatively connected to the at least one processor; wherein the memory 012 stores instructions that can be executed by the at least one processor 011, and the instructions are executed by the at least one processor 011 so that the at least one processor 011 can execute the pipeline defect classification and identification method provided in the embodiment of the present application.

[0037] Figure 2 A schematic diagram of a pipeline defect classification and identification method provided in an embodiment of the present application. Figure 2 As shown, in some embodiments, the pipeline defect classification and identification method includes the following steps:

[0038] S1, obtaining pipeline defect data to be trained.

[0039] In some embodiments, pipeline defect data to be trained is obtained based on two pipeline pull-through experiments. Figure 3 A schematic diagram of the structure of a robot that obtains pipeline defect data to be trained in a pipeline defect classification and recognition method provided in an embodiment of the present application. Figure 3 As shown, in one implementation, the robot may include a leakage magnetic joint, a universal joint, and an electronic package joint. The robot performs a pipeline pulling experiment based on the magnetic flux theory. The purpose of the pulling experiment is to find defects in the ferromagnetic material being tested by measuring the magnetic flux leaking from the surface of the magnetized material. The experiment was repeated twice to ensure the reliability of the experimental data. It can be understood that the two pulling data are complete and the defect signal has good repeatability. After verifying that the pulling speed has no significant effect on the detection data, the detection data is identified and classified.

[0040] In some embodiments, the pipeline defect data to be trained includes a training set and a test set, and both the training set and the test set include at least part of the pipeline defect data to be trained and a plurality of images without pipeline defects.

[0041] S2, based on the pipeline defect data to be trained, train the pre-trained model to obtain a feature extractor.

[0042] The pre-trained model can be a Vision Transformer (ViT), which is a deep learning model architecture used to process image data and capture the global information of the image through a self-attention mechanism. Pipeline defect data is identified and the performance of pre-trained learning algorithms that can be used for object detection and image classification is recorded.

[0043] Figure 4 This is a schematic diagram of the structure of a pre-trained model in a pipeline defect classification and identification method provided in an embodiment of the present application. Figure 4 As shown, in some embodiments, when executing step S2, the pipeline defect classification and identification method further includes:

[0044] S21, inputting the pipeline defect data to be trained into a block linear projection module to obtain multiple data blocks.

[0045] Linear Projection of Flattened Patches: Linearly project the image sequence after being flattened to obtain a higher-dimensional feature representation.

[0046] Figure 5 Four example images of defect images input into a pre-trained model in a pipeline defect classification and recognition method provided in an embodiment of the present application. Figure 5 As shown, the defect image is input into the block linear projection module.

[0047] S22, preprocessing the data block based on image segmentation, position embedding and category embedding to obtain a preprocessed data block.

[0048] Image patching (Patch), which divides the input image into smaller image patches for input into the visual transformer model.

[0049] Position Embeddings are used to represent the embedding vectors of different positions in the image so that the model can understand the position information of different areas in the image.

[0050] Class Embedding: An embedding vector representing the image category for category prediction in the classifier transformer.

[0051] S23, input the pre-processed data block into the converter encoder.

[0052] Transformer Encoder, which is used to encode the input sequence and extract features.

[0053] In some embodiments, when executing step S23, the pipeline defect classification and identification method further includes:

[0054] S230, training the preprocessed data block based on the embedded image block, normalization, multi-head attention, normalization and multi-layer perceptron in sequence.

[0055] Embedded Patches: The process of dividing an image into multiple patches and then converting each patch into an embedded vector.

[0056] Normalization (Norm), in deep learning models, normalization is often used to limit the numerical range of features or hidden representations to a certain range to reduce instability in model training.

[0057] Multi-Head Attention, an attention mechanism that splits the attention calculation into multiple heads to increase the model's ability to pay attention to different features and positions.

[0058] Multilayer Perceptron (MLP) is a common artificial neural network (ANN) architecture. It consists of multiple neural network layers, each of which is fully connected to the next layer.

[0059] In some embodiments, when executing step S230, the pipeline defect classification and identification method further includes:

[0060] S231, input the preprocessed data block into the embedded image block to obtain a first data result.

[0061] S232, normalize the first data result and perform multi-head attention in sequence to obtain the second data result.

[0062] S233, combining the first data result and the second data result to obtain a third data result.

[0063] S234, normalize the third data result and input it into a multi-layer perceptron to obtain a fourth data result.

[0064] S235, combining the third data result and the fourth data result to obtain a fifth data result.

[0065] Used as Figure 4 The pre-trained ViT encoder structure for pipeline defect classification shown in the figure implements the computer vision algorithm Cascade ViT for pipeline defect classification. Except for the last layer, we retain almost the same architecture as the original ViT, including layers and weights. During the hyperparameter fine-tuning process, the pre-trained model is used as a feature extractor, and a new classifier is added on it to train the new dataset.

[0066] Intersection over Union (IoU) is one of the commonly used evaluation indicators in object detection. It is used to measure the degree of overlap between two bounding boxes. In one implementation, the IoU threshold is set to 0.5. The IoU threshold refers to the threshold used to determine whether a predicted box matches the ground truth box in the object detection task. Usually, when the IoU between a predicted box and a ground truth box is greater than or equal to the set threshold, the predicted box will be regarded as a correct detection result.

[0067] Figure 6 This is a loss or accuracy curve diagram of the training and verification process in a pipeline defect classification and identification method provided in an embodiment of the present application. Figure 6 As shown, in one implementation, 100 normal images without any defects are added to the dataset, and the IoU threshold is set to about 0.5 for the first stage of the proposed method, where the training or validation dataset should contain images without pipeline defects so that the model can learn all relevant features, and this method can effectively prevent noisy normal data from being misclassified as abnormal during the classification process.

[0068] S3, training a classification recognition model based on the feature extractor and the classifier and the training data set. In some embodiments, after executing step S3, the pipeline defect classification recognition method further includes:

[0069] S4, constructs a confusion matrix to evaluate the classification accuracy of the classification recognition model.

[0070] Figure 7 A schematic diagram of a confusion matrix diagram in a pipeline defect classification and identification method provided in an embodiment of the present application. In some embodiments, the confusion matrix indicates the performance of the pipeline defect classification pre-trained model. The labels from 0 to 7 correspond to the indexes and defect types listed in Table 1, and label 8 is the added defect-free image. The pre-trained model captures all pipeline defects well, with a prediction accuracy of 96.75%.

[0071] Figure 8 This is a graph showing the accuracy of different types of defects identified in a pipeline defect classification and identification method provided in an embodiment of the present application. Figure 8 As shown, the accuracy of different types of defect recognition.

[0072] Fig. 9 is a performance graph of different deep learning architectures. Fig. 9As shown, the trained ViT model outperforms other models in terms of precision, recall, and F1 score, indicating that this method can be applied to MFL data analysis for pipeline defect detection and identification with better performance.

[0073] Fig.10 This is a comparison chart of different deep learning architectures. Fig.10 As shown in Figure 2, the defect classifications produced by the multi-class YOLOv5 model and the ViT model are compared with the manual labels from the original dataset. The bounding boxes predicted by YOLOv5 align well with the manually labeled boxes, allowing them to be extracted and input into the ViT model for defect identification. Although the multi-class YOLOv5 model outputs classification results ( Fig.10 b), but its predicted labels are inaccurate compared to the original data ( Fig.10 a). Yolov5 is fast, accurate, and small when selecting object recognition boxes, with a recognition accuracy of nearly 100% on existing datasets. However, Yolov5 performs significantly worse than VIT on classification problems. In contrast, the pre-trained ViT model produces correct predictions. The pre-trained ViT model identifies the type of defect with significantly higher average precision and recall, achieving 30% and 60% improvements respectively. The ViT model also outperforms other popular image classification algorithms. It runs relatively fast, with defect recognition taking only a few seconds for a single image.

[0074] Compared with the single-stage target detection algorithm, the pipeline defect classification and identification method provided in the embodiment of the present application balances the accuracy and efficiency of defect detection, has advantages in small sample prediction problems, and can save model training costs. In addition, it can be easily extended to further quantification or three-dimensional reconstruction of pipeline defects, laying the foundation for further research.

[0075] Based on the same application concept, a pipeline defect classification and identification system is also provided in an embodiment of the present application. The method corresponding to the pipeline defect classification and identification system may be the pipeline defect classification and identification method in the aforementioned embodiment, and its principle of solving the problem is similar to that of the method. The pipeline defect classification and identification system provided in an embodiment of the present application includes at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the methods and / or technical solutions of the aforementioned multiple embodiments of the present application.

[0076] Another embodiment of the present application further provides a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the methods and / or technical solutions of any one or more embodiments of the present application described above.

[0077] Specifically, the present embodiment may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device, or device.

[0078] Computer readable signal media may include a data signal propagated in baseband or as part of a carrier wave, which carries a computer readable program code. Such propagated data signals may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than a computer readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0079] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0080] Computer program code for performing the operation of the present application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., using an Internet service provider to connect through the Internet).

[0081] The secondary flow chart or block diagram in the accompanying drawings shows the possible architecture, function and operation of the equipment, method and computer program product according to various embodiments of the present application. In this regard, each square box in the secondary flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or the secondary flow chart, and the combination of the square boxes in the block diagram and / or the secondary flow chart can be implemented with a dedicated system for hardware that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0082] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0083] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, 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. For example, multiple units or page 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 an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0084] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0085] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0086] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform some steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program codes.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

[0088] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in a device claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

Claims

1. A pipeline defect classification and identification method, characterized in that: include: Obtain pipeline defect data to be trained; Based on the pipeline defect data to be trained, training a pre-trained model to obtain a feature extractor; According to the feature extractor and the classifier, a classification recognition model is trained based on a training data set.

2. The pipeline defect classification and identification method according to claim 1 is characterized in that: The pipeline defect data to be trained includes a training set and a test set, and both the training set and the test set include at least part of the pipeline defect data to be trained and a plurality of images without pipeline defects.

3. The pipeline defect classification and identification method according to claim 1 or 2, characterized in that: When executing the step of training the pre-trained model based on the pipeline defect data to be trained to obtain a feature extractor, the pipeline defect classification and identification method further includes: Inputting the pipeline defect data to be trained into a block linear projection module to obtain multiple data blocks; Preprocessing the data block based on image segmentation, position embedding and category embedding to obtain a preprocessed data block; The pre-processed data blocks are input to a transformer encoder.

4. The pipeline defect classification and identification method according to claim 3 is characterized in that: When executing the step of inputting the pre-processed data block into the converter encoder, the pipeline defect classification and identification method further includes: The preprocessed data block is trained based on embedded image patches, normalization, multi-head attention, normalization, and multi-layer perceptron in sequence.

5. The pipeline defect classification and identification method according to claim 4 is characterized in that: When performing the steps of training the preprocessed data block based on the embedded image block, normalization, multi-head attention, normalization and multi-layer perceptron in sequence, the pipeline defect classification and identification method further includes: Inputting the preprocessed data block into the embedded image block to obtain a first data result; The first data results are normalized and multi-headedly paid attention to obtain the second data results; Combining the first data result and the second data result to obtain a third data result; Normalizing the third data result and inputting it into a multilayer perceptron to obtain a fourth data result; The third data result and the fourth data result are combined to obtain a fifth data result.

6. The pipeline defect classification and identification method according to claim 1 or 2, characterized in that: When executing the step of obtaining pipeline defect data to be trained, the pipeline defect classification and identification method further includes: Based on two pipeline pull-through experiments, the pipeline defect data to be trained is obtained.

7. The pipeline defect classification and identification method according to claim 1 or 2, characterized in that: The intersection-over-union ratio threshold is set to 0.

5.

8. The pipeline defect classification and identification method according to claim 1 or 2, characterized in that: The pipeline defect classification and identification method further includes: A confusion matrix is ​​constructed to evaluate the classification accuracy of the classification recognition model.

9. A pipeline defect classification and identification system, characterized in that: include: at least one processor; a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method according to any one of claims 1 to 8.

10. A computer readable medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the method according to any one of claims 1 to 8.

Citation Information

Patent Citations

  • Oil and gas pipeline corrosion detection method based on magnetic stress

    CN116309580A

  • Construction method and application of defect target detection model of industrial defect detection scene

    CN116777902A

  • Pipeline weld defect detection method based on Swin Transform

    CN117388256A

  • Probability of detection of lifecycle phases of corrosion under insulation using artificial intelligence and temporal thermography

    WO2024035640A2