Construction method for fault grouping model, and fault grouping method and system

Through the training cycle generation of adversarial networks and multi-scale attention networks, the problems of insufficient fault samples and inaccurate labeling are solved, the accuracy of fault grouping models is improved, and the effective identification of fault features is achieved.

WO2024239581A1PCT designated stage expired Publication Date: 2024-11-28CHINA NAT PETROLEUM CORP +2
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2023/135847
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-22
Filing Date
2023-12-01
Publication Date
2024-11-28

AI Technical Summary

Technical Problem

The existing fault grouping model has problems such as insufficient fault samples and inaccurate labeling in seismic exploration, resulting in poor fault feature recognition effect.

Method used

Through the training loop, an adversarial network is generated, and a first-style image data set composed of multiple straight lines and a second-style image data set composed of tomographic style is generated, and annotated using clustering ideas, combined with multi-scale attention Force network improves the accuracy of the fault grouping model.

Benefits of technology

Provide sufficient fault sample data, improve the accuracy of fault grouping model, reduce the subjectivity of manual labeling, and enhance the ability to identify fault features.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2023135847_28112024_PF_FP_ABST
    Figure CN2023135847_28112024_PF_FP_ABST
Patent Text Reader

Abstract

The present invention belongs to the field of fault diagnosis. Provided are a construction method for a fault grouping model, and a fault grouping method and system. The construction method for a fault grouping model comprises: training a cyclic generative adversarial network by using a first style image dataset consisting of a plurality of straight lines and a second style image dataset consisting of faults, so as to obtain an image sample having a fault style, wherein the cyclic generative adversarial network comprises a first generator, a second generator, a first discriminator and a second discriminator, the first generator being used for generating a fault image from a first style image, the second generator being used for generating a straight line image from a second style image, and the first discriminator and the second discriminator being used for determining the probability that an input image is a real image; performing fault tag labeling on the image sample to obtain an image sample having a fault tag; and training a fault grouping neural network by means of the image sample having the fault tag, so as to obtain a fault grouping model. The embodiments of the present invention can improve the accuracy of the fault grouping model.
Need to check novelty before this filing date? Find Prior Art

Description

A method for constructing a fault grouping model, a fault grouping method and a system

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of Chinese patent application No. 202310579001.5 filed on May 22, 2023, the contents of which are incorporated herein by reference. Technical Field

[0003] The present invention relates to the field of fault diagnosis, and in particular to a method for constructing a fault grouping model, a fault grouping method and a system. Background Art

[0004] Fault identification plays an important role in seismic exploration and is an indispensable part of seismic exploration.

[0005] Due to differences in the work area and geological structure, seismic data varies significantly, resulting in irregular distribution of faults. Furthermore, manual fault labeling is subjective and prone to incomplete or inaccurate labeling. Consequently, existing fault grouping models suffer from insufficient fault samples and inaccurate fault labeling, making them incapable of effectively identifying fault characteristics.

[0006] Summary of the Invention

[0007] The purpose of the embodiments of the present invention is to provide a method for constructing a fault grouping model, a fault grouping method and a system. The embodiments of the present invention can solve or partially solve the problems existing in the prior art.

[0008] In order to achieve the above-mentioned purpose, an embodiment of the present invention provides a method for constructing a fault grouping model, which comprises: using a first style image dataset consisting of multiple straight lines and a second style image dataset consisting of faults to train a cyclic generative adversarial network to obtain image samples with a fault style; wherein the cyclic generative adversarial network comprises a first generator, a second generator, a first discriminator and a second discriminator; the first generator is used to generate the first style image as a fault image, and the second generator is used to generate the second style image as a straight line image; the first discriminator and the second discriminator are used to determine whether the input image is an image generated by the generator or an image from the first style image dataset / the second style image dataset; the image samples are labeled with fault labels to obtain image samples with fault labels; and the fault grouping neural network is trained by the image samples with fault labels to obtain a fault grouping model.

[0009] Optionally, the method uses a first style image dataset composed of multiple straight lines and a second style image dataset composed of faults to train a cyclic generative adversarial network to obtain image samples with a fault style, including: inputting the first style image dataset composed of multiple straight lines and the second style image dataset composed of faults into the cyclic generative adversarial network to generate a weighted loss function; adjusting the weights of the cyclic generative adversarial network until the weighted loss function value is minimized; using the weights corresponding to the minimum weighted loss function value as optimal parameters to obtain an optimal cyclic generative adversarial network; inputting the first style image dataset composed of multiple straight lines into the optimal cyclic generative adversarial network to obtain image samples with a fault style.

[0010] Optionally, the first style image dataset composed of multiple straight lines and the second style image dataset composed of faults are input into a recurrent generative adversarial network to generate a weighted loss function, including: inputting the first style image into the first generator to generate a fault image; using the first discriminator to judge the probability value of the generated fault image being the second style image, and obtaining the loss function of the first discriminator; inputting the second style image into the second generator to generate a straight line image; using the second discriminator to judge the probability value of the generated straight line image being the first style image, and obtaining the loss function of the second discriminator; inputting the generated straight line image and fault image into the first generator and the second generator respectively to obtain regenerated fault images and straight line images; calculating the Euclidean distance between the regenerated straight line image and fault image and the first style image and the second style image to obtain a third loss function; and using the sum of the loss function of the first discriminator, the loss function of the second discriminator and the third loss function as the weighted loss function.

[0011] Optionally, the step of inputting the first style image into the first generator to generate a tomographic image, and using the first discriminator to determine the probability value of the generated tomographic image being the second style image, and obtaining the loss function of the first discriminator, includes: inputting the first style image x composed of multiple straight lines into the first generator G to generate a tomographic image G(x); inputting the second style image composed of toms and the generated tomographic image G(x) into the first discriminator D Y , output the probability value D that the tomographic image is the second style image Y (G(x)), we get the loss function of the first discriminator: L GAN (G,D y ,X,Y)=αE y~pdata(y) [log D Y (y)]+(1-α)E x~pdata(x) [log(1--D Y (G(x)))]

[0012] Among them, the p data(x) is the data distribution of the first style image dataset consisting of multiple straight lines, p data(y) is the data distribution of the second style image dataset composed of faults, and α is a preset parameter.

[0013] Optionally, the step of inputting the second style image into the second generator to generate a straight line image, and using the second discriminator to determine the probability value of the generated straight line image being the first style image, and obtaining the loss function of the second discriminator includes: inputting the second style image y composed of faults into the second generator F to generate a straight line image F(y); inputting the first style image composed of multiple straight lines and the generated straight line image F(y) into the second discriminator D X , output the probability value D that the straight line image is the first style image X (F(y)), we get the loss function of the second discriminator: L GAN (G, D X , X, Y) = αE x~pdata(x) [logD X (x)]+(1-α)E y~pdata(y) [log(1-D X (F(y)))]

[0014] Among them, α is a pre-set adjustment factor.

[0015] Optionally, the step of inputting the generated straight line image and tomographic image into the first generator and the second generator respectively to obtain regenerated tomographic image and straight line image, and calculating the Euclidean distance between the regenerated straight line image and tomographic image and the first style image and the second style image to obtain a third loss function includes: inputting the generated straight line image F(y) and tomographic image G(x) into the first generator G and the second generator F respectively to obtain regenerated tomographic image G(F(y)) and straight line image F(G(x)); calculating the Euclidean distance between the regenerated tomographic image G(F(y)) and straight line image F(G(x)) and the second style image y and the first style image x to obtain the third loss function: L cyc (G,F)=E x~pdata(x) [||F(G(x))-x||1]E y~pdata(y) [||G(F(y))-y||1]

[0016] Optionally, the step of labeling the image sample with a slice label to obtain the image sample with the slice label includes calculating the unknown pixel point x of the image sample. i and the first style image known pixel u jThe K known pixels with the smallest distance are taken as the nearest neighboring samples; the categories to which the majority of the K nearest neighboring samples belong are taken as the categories of the unknown pixels, and the fault labels are annotated.

[0017] Optionally, the unknown pixel x of the image sample is calculated using the following formula: i and the first style image known pixel u j The distance dist:

[0018] Where f is the Gaussian function and e is the norm, which takes a value of 1 or 2.

[0019] Optionally, the fault grouping neural network is a YOLACT network, including: a feature extraction network, a mask coefficient network, a prototype network, a multi-scale attention network, and an aggregation and trimming network; the feature extraction network is used to extract the features of the tomographic image and form a feature map; the mask coefficient network is used to process the feature map to obtain a prediction box, classification and mask coefficients; the multi-scale attention network is used to weight the feature map to obtain a weighted feature map; the prototype network is used to perform feature prediction based on the weighted feature map and generate a prototype; the aggregation and trimming network is used to combine the prediction box, classification and mask coefficients and prototype to identify faults.

[0020] Optionally, weighting the feature map to obtain a weighted feature map includes: compressing the feature map to obtain a compressed vector; reducing the dimension of the compressed vector through a convolution layer to obtain an attention weight; and multiplying the attention weight by the feature map to obtain a weighted feature map.

[0021] Correspondingly, an embodiment of the present invention further provides a fault grouping method, which includes: establishing a fault grouping model using the fault grouping model construction method; inputting the image to be identified into the fault grouping model to obtain a fault grouping result.

[0022] Correspondingly, an embodiment of the present invention further provides a system for constructing a fault grouping model, which is used to execute the method for constructing a fault grouping model.

[0023] Correspondingly, an embodiment of the present invention further provides a fault grouping system for executing the fault grouping method.

[0024] In an embodiment of the present invention, by training a cyclic generative adversarial network using a first style image dataset consisting of multiple straight lines and a second style image dataset consisting of faults, image samples with a fault style can be obtained, thereby providing sufficient fault sample data for training the fault grouping neural network and improving the accuracy of the fault grouping model.

[0025] Furthermore, the embodiment of the present invention utilizes the concept of clustering to first calculate the distance between the unknown pixel points of the image sample and the known pixel points of the first style image, and then uses the category of the majority of the nearest samples as the category of the unknown pixel point for fault labeling. This solves the problems of low labeling efficiency, training difficulties, and poor result accuracy in existing technologies.

[0026] Furthermore, the fault grouping neural network provided by the embodiment of the present invention includes a multi-scale attention network. By weighting the feature map, the importance of different features in the feature map to the segmentation accuracy can be distinguished, thereby improving the accuracy of fault grouping.

[0027] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following detailed description, they are used to explain the embodiments of the present invention, but do not constitute a limitation of the embodiments of the present invention. In the accompanying drawings:

[0029] FIG1 is a flow chart of a method for constructing a fault grouping model provided by an embodiment of the present invention;

[0030] FIG2 is a flow chart of a method for training a cyclic generative adversarial network according to an embodiment of the present invention;

[0031] FIG3 is a structural diagram of a cyclic generative adversarial network provided by an embodiment of the present invention;

[0032] FIG4 is a diagram of a YOLACT network structure provided by an embodiment of the present invention;

[0033] FIG5 is a structural diagram of a multi-scale attention network provided by an embodiment of the present invention;

[0034] FIG6 is a flow chart of a fault grouping method provided by an embodiment of the present invention;

[0035] FIG7 is a first style image composed of multiple straight lines provided by an embodiment of the present invention;

[0036] FIG8 is an image with a tomographic style generated by a cyclic generative adversarial network provided by an embodiment of the present invention;

[0037] FIG9 is a fragmentation-style image after data processing provided by an embodiment of the present invention;

[0038] FIG10 is a diagram showing the slice grouping results provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The following describes the specific implementation of the embodiment of the present invention in detail with reference to the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the embodiment of the present invention and is not used to limit the embodiment of the present invention.

[0040] FIG1 is a flow chart of a method for constructing a fault grouping model provided by an embodiment of the present invention.

[0041] As shown in FIG1 , the method for constructing the fault grouping model includes:

[0042] S110: Using a first style image dataset consisting of multiple straight lines and a second style image dataset consisting of faults to train a cyclic generative adversarial network to obtain image samples with a fault style; wherein the cyclic generative adversarial network includes a first generator, a second generator, a first discriminator, and a second discriminator, the first generator is used to generate the first style image as a fault image, the second generator is used to generate the second style image as a straight line image, and the first discriminator and the second discriminator are used to determine whether the input image is an image generated by the generator or an image from the first style image dataset / the second style image dataset;

[0043] S120: labeling the image sample with a slice label to obtain an image sample with a slice label;

[0044] S130: Training a slice grouping neural network using the image samples with slice labels to obtain a slice grouping model.

[0045] In step S110, a first style image dataset consisting of multiple straight lines and a second style image dataset consisting of faults can be used to train a cyclic generative adversarial network, and then the cyclic generative adversarial network can be used to obtain image samples with fault style, thereby meeting the requirements for fault samples when constructing a fault grouping model.

[0046] The first style image dataset consisting of multiple straight lines can be obtained by automatically generating a plurality of straight lines on an image, and the second style image dataset consisting of faults can be obtained by randomly cropping an existing fault image.

[0047] Furthermore, when using a first style image dataset composed of multiple straight lines and a second style image dataset composed of faults to train a cyclic generative adversarial network, the first style image dataset composed of multiple straight lines and the second style image dataset composed of faults can be input into the cyclic generative adversarial network to generate a weighted loss function; then, the weights of the cyclic generative adversarial network are adjusted until the weighted loss function value is minimized; the weights corresponding to the minimum weighted loss function value are used as the optimal parameters to obtain the optimal cyclic generative adversarial network; the first style image dataset composed of multiple straight lines is input into the optimal cyclic generative adversarial network to obtain image samples with fault style.

[0048] It should be noted that the smaller the total loss value of the weighted loss function is, the better the training effect of the network is, and the style of the generated image is closer to the style of the tomographic image.

[0049] For ease of understanding, an embodiment of the present invention provides a flowchart of a method for training a cyclic generative adversarial network, as shown in FIG2 . The method for training a cyclic generative adversarial network includes:

[0050] S210: Inputting a first style image dataset consisting of a plurality of straight lines and a second style image dataset consisting of faults into a recurrent generative adversarial network to generate a weighted loss function;

[0051] S220: Adjusting the weights of the cyclic generative adversarial network until the weighted loss function value is minimized;

[0052] S230: Taking the weight value corresponding to the minimum value of the weighted loss function as the optimal parameter to obtain the optimal recurrent generative adversarial network;

[0053] S240: Inputting the first style image dataset consisting of multiple straight lines into the optimal cycle generative adversarial network to obtain image samples with a fault style.

[0054] The following is a detailed introduction to the method of obtaining the weighted loss function during the training cycle of the generative adversarial network in step S210.

[0055] FIG3 is a structural diagram of a cyclic generative adversarial network provided by an embodiment of the present invention.

[0056] As shown in FIG3 , first, the first style image is input to the first generator (generator G) to generate a tomographic image; and the first discriminator (discriminator D) is used to generate a tomographic image. Y ) determines the probability value of the generated tomographic image being the second style image, and obtains the loss function of the first discriminator; then, the second style image is input to the second generator (generator F) to generate a straight line image; using the second discriminator (discriminator D X) Determine the probability value of the generated straight line image being the first style image, and obtain the loss function of the second discriminator; secondly, input the generated straight line image and tomographic image into the first generator and the second generator respectively, and obtain the regenerated tomographic image and straight line image; calculate the Euclidean distance between the regenerated straight line image and tomographic image and the first style image and the second style image, and obtain the third loss function; finally, the sum of the loss function of the first discriminator, the loss function of the second discriminator and the third loss function is used as the weighted loss function.

[0057] For example, the first style image x composed of multiple straight lines can be input into the first generator G to generate a tomographic image G(x); the second style image composed of toms and the generated tomographic image G(x) can be input into the first discriminator D Y , output the probability value D that the tomographic image is the second style image Y (G(x)), we get the loss function of the first discriminator: L GAN (G, D Y , X, Y) = αE y~pdata(y) [log D Y (y)]+(1-α)E x~pdata(x) [log(1-D Y (G(x)))] wherein the p data(x) is the data distribution of the first style image dataset consisting of multiple straight lines, p data(y) is the data distribution of the second style image dataset composed of faults, and α is a pre-given adjustment factor.

[0058] Then, the second style image y composed of faults is input into the second generator F to generate a straight line image F(y); the first style image composed of multiple straight lines and the generated straight line image F(y) are input into the second discriminator D X , output the probability value D that the straight line image is the first style image X (F(y)), we get the loss function of the second discriminator: L GAN (G,D X ,X,Y)=αE x~pdata(x) [logD X (x)]+(1-α)E y~pdata(y) [log(1-D X (F(y)))],

[0059] Among them, α is a pre-given adjustment factor.

[0060] Furthermore, the generated straight line image F(y) and tomographic image G(x) are respectively inputted into the first generator G and the second generator F to obtain the regenerated tomographic image G(F(y)) and straight line image F(G(x));

[0061] Calculate the Euclidean distance between the regenerated tomographic image G(F(y)), the straight line image F(G(x)), the second style image y, and the first style image x to obtain the third loss function: L cyc (G,F)=E x~pdata(x) [||F(G(x))-x||1]+E y~pdata(y) [||G(F(y))-y||1]

[0062] Finally, the sum of the loss function of the first discriminator, the loss function of the second discriminator and the third loss function is used as the weighted loss function: L(G,F,D X ,D Y )=L GAN (G,D Y ,X,Y)+L GAN (F,D X ,Y,X)+L cyc (G,F).

[0063] In an embodiment of the present invention, by training a cyclic generative adversarial network using a first style image dataset composed of multiple straight lines and a second style image dataset composed of faults, image samples with fault styles can be obtained, thereby providing sufficient sample data for training the fault grouping neural network and improving the accuracy of the fault grouping model.

[0064] In step S120, for the line image samples in the first style image dataset, which consist of multiple lines, the category to which each pixel belongs is known. However, the pixels in the tomographic image obtained after tomographic style transfer are unknown samples. Therefore, when labeling these image samples, clustering can be used to classify the unknown samples, reducing the subjectivity of manual labeling and improving the accuracy of tomographic grouping.

[0065] Specifically, if most of the samples that are most similar to a sample in the feature space (i.e., the samples with the closest distance in the feature space) belong to a certain category, then the sample also belongs to this category.

[0066] The following introduces a method for labeling the image samples using clustering ideas.

[0067] First, the unknown pixel x of the image sample can be calculated i and the first style image known pixel u j The distance dist: Among them, f is a Gaussian function, e is a norm, which can take values ​​such as 1 and 2; then, the K known pixels with the smallest distance are taken as the nearest neighboring samples; finally, the category to which the majority of the K nearest neighboring samples belong is taken as the category of the unknown pixel point, and the fault label is annotated.

[0068] For example, first, we can calculate the unknown pixel point (point to be classified) x in the tomographic image i and the known pixel u in the straight line image j The distance dist: Where f is the Gaussian function and e is the norm, which can take values ​​such as 1 and 2. The maximum distance max_dist among the current K most similar pixels is obtained. If dist is less than max_dist, the pixel is used as the K nearest neighbor samples.

[0069] Repeat the above steps until the distance calculation between the unknown pixel and all known pixels is completed; then, count the number of times each category appears in the K nearest neighbor samples, and select the category with the highest frequency as the category of the unknown pixel.

[0070] The embodiment of the present invention utilizes the idea of ​​clustering. First, the distance between the unknown pixel points of the image sample and the known pixel points of the first style image is calculated. The category to which the majority of the nearest samples belong is used as the category of the unknown pixel points for fault labeling. This solves the problems of low labeling efficiency, training difficulties, and poor result accuracy in the existing technology.

[0071] In step S130, the fault grouping neural network may be a YOLACT network. FIG4 is a diagram of the YOLACT network structure provided by an embodiment of the present invention.

[0072] As shown in Figure 4, the YOLACT network may include: a feature extraction network, a mask coefficient network, a prototype network, a multi-scale attention network, and an aggregation and pruning network.

[0073] Among them, the feature extraction network is used to extract the features of the tomographic image and form a feature map; the mask coefficient network is used to process the feature map to obtain a prediction box, classification and mask coefficients; the multi-scale attention network is used to weight the feature map to obtain a weighted feature map; the prototype network is used to perform feature prediction based on the weighted feature map and generate a prototype; the aggregation and trimming network is used to combine the prediction box, classification and mask coefficients and prototype to identify the fault.

[0074] The YOLACT network in the prior art ignores the features between feature images of different scales, as well as the importance of different features to segmentation accuracy, resulting in low image segmentation accuracy. Therefore, the improved YOLACT network provided in an embodiment of the present invention includes a multi-scale attention network. Figure 5 is a structural diagram of the multi-scale attention network provided in an embodiment of the present invention. As shown in Figure 5, when the multi-scale attention network is used to weight the feature map to obtain a weighted feature map, the feature map can be first compressed to obtain a compressed vector; then, the compressed vector is reduced in dimension through a convolutional layer to obtain an attention weight; and the attention weight is multiplied by the feature map to obtain a weighted feature map, thereby reflecting the features between feature images of different scales and the different importance of different features to segmentation accuracy.

[0075] For example, when the multi-scale attention network is used to weight the feature map to obtain the weighted feature map, it first obtains feature images of different scales by cascading 1×1 convolution modules with different holes and global average pooling operations, and fuses them in a channel splicing manner; then, the input of the channel attention network is the feature map f∈R after multi-scale fusion. H×W×C , where H, W, and C represent the height, width, and number of input channels, respectively;

[0076] Afterwards, the feature map is compressed using the global average pooling operation, and the compressed vector g∈R is output. 1×1×C , the calculation formula is as follows: Where: c=1,2,…,C;

[0077] Furthermore, the vector obtained above is reduced in dimension through a convolutional layer, and then passed through a sigmoid activation function to obtain the attention weight: ω c =σ(w(g c ))(W represents a 1×1 convolutional layer, σ represents a sigmoid activation function, ω c Weighs the cth feature channel f c feature importance).

[0078] Finally, the above weight ω is multiplied by the input feature map f to weight the feature map and obtain the weighted feature image:

[0079] The fault grouping neural network provided by the embodiment of the present invention includes a multi-scale attention network. By weighting the feature map, a weighted feature map can be obtained to distinguish the importance of different features in the feature map for segmentation accuracy, thereby improving the accuracy of fault grouping.

[0080] Correspondingly, an embodiment of the present invention also provides a fault grouping method.

[0081] FIG6 is a flow chart of a fault grouping method provided by an embodiment of the present invention. As shown in FIG6 , the fault grouping method includes:

[0082] S610: Establishing a fault grouping model using the fault grouping model construction method;

[0083] S620: Input the image to be identified into the slice grouping model to obtain a slice grouping result.

[0084] In a specific embodiment, a first style image dataset consisting of multiple straight lines can be obtained by automatically generating a plurality of straight lines on an image, and a second style image dataset consisting of faults can be obtained by randomly cropping an existing fault image to train a cyclic generative adversarial network;

[0085] Specifically, the first style image is input into the first generator to generate a tomographic image; the first discriminator is used to determine a probability value of the generated tomographic image being a second style image, and a loss function of the first discriminator is obtained;

[0086] Inputting the second style image into the second generator to generate a straight line image; using the second discriminator to determine the probability value of the generated straight line image being the first style image, and obtaining a loss function of the second discriminator;

[0087] inputting the generated straight line image and tomographic image into the first generator and the second generator respectively to obtain the regenerated tomographic image and straight line image;

[0088] Calculating the Euclidean distance between the regenerated tomographic image, the straight line image, and the first style image to obtain a third loss function; and using the sum of the loss function of the first discriminator, the loss function of the second discriminator, and the third loss function as a weighted loss function;

[0089] Adjusting the weights of the recurrent generative adversarial network until the weighted loss function value is minimized; using the weights corresponding to the minimum weighted loss function value as optimal parameters to obtain an optimal recurrent generative adversarial network; inputting the first style image dataset consisting of multiple straight lines into the optimal recurrent generative adversarial network to obtain image samples with a fault style;

[0090] Then, we use a clustering algorithm to add annotations to the tomographic images and obtain image samples with tomographic labels. The image samples with tomographic labels are divided into a training set and a test set in a ratio of 8:2.

[0091] Afterwards, the image samples (training set) are input into the improved YOLACT network for training and optimization to obtain the fault grouping model.

[0092] The data of the actual work area are input into the fault grouping model to obtain the fault grouping results.

[0093] Among them, Figure 7 is a first style image composed of multiple straight lines provided by an embodiment of the present invention; Figure 8 is an image sample with a fault style generated by a cyclic generative adversarial network provided by an embodiment of the present invention; Figure 9 is an image sample after data processing provided by an embodiment of the present invention; Figure 10 is a fault grouping result diagram provided by an embodiment of the present invention, where labels 1, 2, and 3 represent different fault lines.

[0094] In an embodiment of the present invention, by training a cyclic generative adversarial network using a first style image dataset consisting of multiple straight lines and a second style image dataset consisting of faults, image samples with a fault style can be obtained, thereby providing sufficient fault sample data for training the fault grouping neural network and improving the accuracy of the fault grouping model.

[0095] Furthermore, the embodiment of the present invention utilizes the concept of clustering to first calculate the distance between the unknown pixel points of the image sample and the known pixel points of the first style image, and then uses the category of the majority of the nearest samples as the category of the unknown pixel point for fault labeling. This solves the problems of low labeling efficiency, training difficulties, and poor result accuracy in existing technologies.

[0096] Furthermore, the fault grouping neural network provided in the embodiment of the present invention includes a multi-scale attention network. By weighting the feature map, a weighted feature map can be obtained to distinguish the different importance of different features in the feature map for segmentation accuracy, thereby improving the accuracy of fault grouping.

[0097] Correspondingly, an embodiment of the present invention further provides a system for constructing a fault grouping model, which is used to execute the method for constructing a fault grouping model.

[0098] Correspondingly, an embodiment of the present invention further provides a fault grouping system for executing the fault grouping method.

[0099] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0100] The present application is described with reference to the flow chart and / or block diagram of the method, device (system), and computer program product according to the embodiment of the present application. It should be understood that each flow process and / or box in the flow chart and / or block diagram and the combination of the flow process and / or box in the flow chart and / or block diagram can be realized by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processing machine or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device for realizing the function specified in one flow chart flow or multiple flows and / or one box or multiple boxes of the block diagram.

[0101] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0102] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0103] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0104] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0105] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0106] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0107] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for constructing a fault grouping model, characterized in that: The method for constructing the fault grouping model comprises: A cyclic generative adversarial network is trained using a first style image dataset consisting of a plurality of straight lines and a second style image dataset consisting of faults to obtain image samples with a fault style; wherein the cyclic generative adversarial network comprises a first generator, a second generator, a first discriminator and a second discriminator, the first generator is used to generate the first style image as a fault image, the second generator is used to generate the second style image as a straight line image, the first discriminator and the second discriminator are used to determine whether the input image is an image generated by the generator or an image from the first style image dataset / the second style image dataset; Performing slice label marking on the image samples to obtain image samples with slice labels; The fault grouping neural network is trained by using the image samples with fault labels to obtain a fault grouping model.

2. The method for constructing a fault grouping model according to claim 1, characterized in that: The method of using a first style image dataset consisting of a plurality of straight lines and a second style image dataset consisting of faults to train a cyclic generative adversarial network to obtain image samples with a fault style includes: Inputting a first style image dataset consisting of a plurality of straight lines and a second style image dataset consisting of faults into a cyclic generative adversarial network to generate a weighted loss function; Adjusting the weights of the cyclic generative adversarial network until the weighted loss function value is minimized; The weight corresponding to the minimum value of the weighted loss function is taken as the optimal parameter to obtain the optimal recurrent generative adversarial network; The first style image data set consisting of a plurality of straight lines is input into the optimal cycle generative adversarial network to obtain image samples with a fault style.

3. The method for constructing a fault grouping model according to claim 2, characterized in that: The step of inputting a first style image dataset consisting of a plurality of straight lines and a second style image dataset consisting of faults into a cyclic generative adversarial network to generate a weighted loss function includes: Inputting the first style image into the first generator to generate a tomographic image; Using the first discriminator to determine the probability value of the generated tomographic image being a second style image, and obtaining a loss function of the first discriminator; Inputting the second style image into a second generator to generate a straight line image; Using the second discriminator to determine the probability value of the generated straight line image being the first style image, and obtaining a loss function of the second discriminator; Inputting the generated straight line image and tomographic image into the first generator and the second generator respectively, to obtain the regenerated tomographic image and straight line image; Calculating the Euclidean distances between the regenerated straight line image, the tomographic image and the first style image and the second style image to obtain a third loss function; The sum of the loss function of the first discriminator, the loss function of the second discriminator and the third loss function is used as a weighted loss function.

4. The method for constructing a fault grouping model according to claim 3, characterized in that: Inputting the first style image into the first generator to generate a tomographic image, using the first discriminator to determine a probability value of the generated tomographic image being a second style image, and obtaining a loss function of the first discriminator, including: Inputting a first style image x composed of a plurality of straight lines into the first generator G to generate a tomographic image G(x); The second style image composed of tomography and the generated tomographic image G(x) are input to the first discriminator D Y , output the probability value D that the tomographic image is the second style image Y (G(x)), and obtain the loss function of the first discriminator L GAN (G,D Y ,X,Y)9αE y~pdata (y)[log D Y (y)]+(1-α)E x~pdata(x) [ log ( 1 - D Y (G(x)))] Among them, the p data(x) is the data distribution of the first style image dataset consisting of multiple straight lines, p data(y) is the data distribution of the second style image data set composed of faults, and α is a preset parameter.

5. The method for constructing a fault grouping model according to claim 3, characterized in that: The step of inputting the second style image into the second generator to generate a straight line image, using the second discriminator to determine the probability value of the generated straight line image being the first style image, and obtaining the loss function of the second discriminator includes: Input the second style image y composed of faults into the second generator F to generate a straight line image F(y); The first style image composed of multiple straight lines and the generated straight line image F(y) are input into the second discriminator D X , output the probability value D that the straight line image is the first style image X (F(y)), and obtain the loss function of the second discriminator: L GAN (G,D X ,X,Y)=αE x~pdata(x) [log D X (x)]+(1-α)E y~pdata(y) [log(1-D X (F(y)))]。 6. The method for constructing a fault grouping model according to claim 3, characterized in that: The step of inputting the generated straight line image and tomographic image into the first generator and the second generator respectively to obtain the regenerated tomographic image and straight line image, and calculating the Euclidean distance between the regenerated straight line image and tomographic image and the first style image and the second style image to obtain a third loss function includes: Input the generated straight line image F(y) and tomographic image G(x) to the first generator G and the second generator F respectively, so as to obtain the regenerated tomographic image G(F(y)) and straight line image F(G(x)); Calculate the Euclidean distance between the regenerated tomographic image G(F(y)), the straight line image F(G(x)) and the second style image y and the first style image x to obtain a third loss function L cyc (G,F)=E x~pdata(x) [||F(G(x))-x||1]+E y~pdata(y) [||G(F(y))-y||1] 7. The method for constructing a fault grouping model according to any one of claims 1 to 6, characterized in that: The step of labeling the image samples with slice labels to obtain the image samples with slice labels includes: Calculate the unknown pixel x of the image sample i With the first style image known pixel u j distance; Take the K known pixels with the smallest distance as the nearest neighboring samples; The categories to which the majority of K nearest neighboring samples belong are taken as the categories of the unknown pixel points, and the fault labels are annotated.

8. The method for constructing a fault grouping model according to claim 7, characterized in that: The unknown pixel x of the image sample is calculated by the following formula i With the first style image known pixel u j The distance dist: Where f is the Gaussian function and e is the norm, which takes a value of 1 or 2.

9. The method for constructing a fault grouping model according to claim 1, characterized in that: The fault grouping neural network is a YOLACT network, including: a feature extraction network, a mask coefficient network, a prototype network, a multi-scale attention network, and an aggregation and trimming network; The feature extraction network is used to extract features of the tomographic image to form a feature map; The mask coefficient network is used to process the feature map to obtain a prediction box, a classification and a mask coefficient; The multi-scale attention network is used to weight the feature map to obtain a weighted feature map; The prototype network is used to perform feature prediction based on the weighted feature map to generate a prototype; The aggregation and pruning network is used to combine the prediction boxes, classification and mask coefficients, and prototypes to identify faults.

10. The method for constructing a fault grouping model according to claim 9, characterized in that: The step of weighting the feature map to obtain a weighted feature map includes: Compressing the feature map to obtain a compressed vector; Passing a convolutional layer to reduce the dimension of the compressed vector to obtain an attention weight; and The attention weight is multiplied by the feature map to obtain a weighted feature map.

11. A fault grouping method, characterized in that: The fault grouping method comprises: Using the method for constructing a fault grouping model according to any one of claims 1 to 10, a fault grouping model is established; The image to be identified is input into the fault grouping model to obtain a fault grouping result.

12. A system for constructing a fault grouping model, characterized in that: Used to execute the method for constructing the fault grouping model described in any one of claims 1-10.

13. A fault grouping system, characterized in that: Used to execute the fault grouping method described in claim 11.

Citation Information

Patent Citations

  • Method for carrying out style migration on calligraphy based on cyclic generative adversarial network

    CN110570346A

  • Improved method based on YOLACT instance segmentation

    CN113160263A

  • Image style migration system and method based on three-branch clustering semantic segmentation

    CN113902613A

  • Training image-to-image translation neural networks

    US20200160113A1