A method and system for identifying defects of catenary dropper structure abnormalities

Abstract

The application provides a catenary dropper structure abnormal defect identification method and system, and relates to the technical field of catenary defect identification. The method comprises the following steps: collecting a catenary component image, and extracting a dropper area from the catenary component image; performing standardization processing on the dropper area through a spatial transformation network STN to obtain a standardized image with a unified posture; inputting the standardized image into an unsupervised anomaly detection network SA-AnoNet with a fusion space attention mechanism, extracting a structure feature, and calculating a feature difference; comparing the feature difference with a preset threshold value, and identifying the defects of the dropper structure abnormality. The application can be trained without defect samples, has strong anti-interference, high standardization degree, and full-automatic operation, can quickly identify defects such as current-carrying ring abnormality, unforcedness, loss, deviation, and height overrun, significantly improves the detection efficiency and accuracy, and is suitable for intelligent operation and maintenance of electrified railway catenary droppers.

Description

Technical Field

[0001] This invention relates to the field of contact wire defect identification technology, and in particular to a method and system for identifying defects in the structure of contact wire droppers. Background Technology

[0002] In the overhead contact system of electrified railways, the dropper is a key load-bearing and current-conducting component between the catenary and the contact wire. Its structural integrity directly determines the current-collecting stability of the pantograph and the safety of train operation. Current specifications impose strict quantitative requirements on the dropper body, clamps, current-carrying rings, crimped pipes, heart-shaped rings, threaded clamps, locking washers, installation angles, adjacent height differences, and electrical clearances.

[0003] During long-term operation, the dropper may develop structural abnormalities such as broken strands, missing / displaced current-carrying rings, lack of force, loose clamps, twisted surfaces, and excessive height, which may lead to potential hazards such as electric arcs, hard spots, and poor current collection.

[0004] In practice, manual inspection or manual analysis of image data collected on the overhead contact line suspension status is commonly used. However, existing inspection methods have the following drawbacks: 1. Manual on-site inspections are inefficient, have insufficient coverage, and pose high safety risks; 2. Manually interpreting the images detected by the overhead contact line suspension status monitoring device (4C) results in a large data volume, fatigue, poor consistency, and insufficient timeliness. 3. Traditional supervised deep learning requires a large number of labeled defect samples, which makes sample acquisition difficult, has weak generalization ability, and cannot identify unknown defect types.

[0005] Therefore, the industry urgently needs a defect identification method and system for catenary dropper structures that requires no defect samples, has strong anti-interference capabilities, high standardization, and is fully automated, in order to solve the existing technical problems. Summary of the Invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art, and proposes a defect identification method and system for abnormal contact wire suspension structure.

[0007] In a first aspect, embodiments of the present invention provide a method for identifying defects in the structure of overhead contact line droppers, including: Acquire images of overhead contact system components and extract the dropper area from the images of the overhead contact system components; The suspension wire region is standardized by using a Spatial Transformation Network (STN) to obtain a standardized image with a uniform posture. The unsupervised anomaly detection network SA-AnoNet, which integrates standardized image input with spatial attention mechanism, extracts structural features and calculates feature differences. The defects of abnormal suspension wire structure are identified by comparing the feature differences with a preset threshold.

[0008] Furthermore, the dropper area is extracted from the images of the overhead contact system components. Specific methods include: Collect a large number of images of catenary components containing droppers, and label the dropper tension cable clamps and dropper contact wire clamps in the images; Train a target detection model, which is used to achieve precise positioning of the wire clamp; The complete dropper region is obtained by using the minimum bounding rectangle or KM weighted bipartite graph matching.

[0009] Furthermore, the complete dropper region is obtained by using minimum bounding rectangle or KM weighted bipartite graph matching. Specific methods include: If the catenary tension cable clamp and catenary contact wire clamp are not detected in the image of the catenary components, then the catenary area is determined to be missing. If there is only one dropper cable clamp and one dropper contact wire clamp in the image of the overhead contact system components, the complete dropper area is obtained by using the minimum bounding rectangle method. If the image of the overhead contact system components contains more than one dropper cable clamp and one dropper contact wire clamp, the weighted bipartite graph minimum matching algorithm is used to match the pairs of dropper cable clamps and contact wire clamps.

[0010] Furthermore, the suspension string region is standardized using a Spatial Transformation Network (STN). Specific methods include: A localization network is used to generate an affine transformation matrix to describe the spatial transformation of the image of the drop wire region; The mesh generator generates a corresponding mesh based on the affine transformation matrix, thereby realizing the spatial transformation of the input image. The sampler uses bilinear interpolation to output a standardized image.

[0011] Furthermore, the formula for the mesh generator is as follows: ; here Representative input image The coordinates of the point Representative output image The most punctuation mark, Represents the affine transformation matrix, where and The coordinates are normalized to -1 to 1.

[0012] Furthermore, the unsupervised anomaly detection network SA-AnoNet is constructed based on an improved Ganomaly, and the specific construction method includes: Skip-Connection is introduced to form AnoNet, which solves the gradient vanishing problem in deep networks and improves reconstruction accuracy; An embedded spatial attention mechanism module is used to perform channel pooling, convolution, and sigmoid activation on the feature map to generate a spatial weight map and extract key structural features.

[0013] Furthermore, the method for calculating feature differences in the unsupervised anomaly detection network SA-AnoNet includes: During the training phase, only normal dropper images are input to learn the standard structural feature distribution; During the inference phase, the distance between the standard structural features and the reconstructed structural features is calculated. The specific calculation formula includes: ; Where Diff(x) is the feature difference value, used to quantify the degree of difference between the input image and its reconstructed image in the feature space; GE(x) is the high-dimensional feature vector extracted by the encoder GE from the input image x; E(G(x)) is the high-dimensional feature vector extracted by the encoder GE after the input image x is reconstructed by encoding and decoding.

[0014] Furthermore, the aforementioned feature differences are compared with a preset threshold to identify defects in the suspension cable structure. Specific methods include: When the feature difference Diff(x) > preset threshold When it is determined that there is an abnormal defect in the dropper structure, the abnormal defect includes at least the missing dropper current-carrying ring, the dropper not being under force, the dropper being missing, and the dropper current-carrying ring being installed in the wrong direction.

[0015] Secondly, this invention also discloses a defect identification system for abnormal overhead contact line dropper structures, comprising: a dropper area extraction module, a standardization module, an anomaly identification module, and a result output module; wherein: The dropper area extraction module is used to acquire images of catenary components and extract the dropper area from the images of the catenary components. The standardization module is used to standardize the dropper region through a spatial transformation network (STN) to obtain a standardized image with a uniform posture. The anomaly detection module is used to integrate standardized image inputs with the unsupervised anomaly detection network SA-AnoNet, which uses a spatial attention mechanism, to extract structural features and calculate feature differences. The result output module is used to compare the feature differences with a preset threshold to identify defects in the suspension wire structure.

[0016] Thirdly, the present invention also discloses an electronic device, comprising: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement any of the recognition methods.

[0017] This invention provides a method for identifying defects in the structure of overhead contact line droppers, comprising: acquiring images of overhead contact line components and extracting dropper regions from these images; standardizing the dropper regions using a Spatial Transformation Network (STN) to obtain standardized images with uniform posture; inputting the standardized images into an unsupervised anomaly detection network (SA-AnoNet) incorporating a spatial attention mechanism to extract structural features and calculate feature differences; and comparing the feature differences with a preset threshold to identify defects in the dropper structure. This invention requires no defect samples for training, exhibits strong anti-interference capabilities, high standardization, and fully automated operation. It can quickly identify defects such as current-carrying loop anomalies, lack of force, missing loops, offsets, and height exceeding limits, significantly improving detection efficiency and accuracy. It is suitable for intelligent operation and maintenance of overhead contact line droppers in electrified railways.

[0018] Compared with the prior art, the present invention has the following beneficial effects: 1. Fully automated process: This invention replaces manual interpretation, significantly reducing labor costs and improving detection timeliness; 2. No defect samples required: This invention addresses the pain points of scarce defect samples and high annotation costs in engineering. 3. Strong anti-interference: This invention employs Spatial Transformation Network (STN) and spatial attention mechanism to effectively resist vibration, illumination, and background changes; 4. High generalizability: This invention can identify a variety of known and unknown structural defects such as abnormal current-carrying rings, lack of force, missing rings, twisted surfaces, offset rings, and excessive height. Attached Figure Description

[0019] Figure 1 A flowchart illustrating a method for identifying defects in the structure of overhead contact line droppers provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the dropper before STN network transformation according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the dropper image after STN network transformation according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the AnoNet network structure according to an embodiment of the present invention; Figure 5 A structural block diagram of a defect identification system for abnormal overhead contact line dropper structure provided in an embodiment of the present invention; Figure 6 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solutions of the present invention, exemplary embodiments of the present invention are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0021] Where there is no conflict, the various embodiments of the present invention and the features thereof may be combined with each other.

[0022] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0023] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used herein, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded. Terms such as “connected” or “linked” are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0024] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art and the invention, and will not be interpreted as having an idealized or overly formal meaning unless expressly so defined herein.

[0025] In the technical solution of this invention, the collection, storage, use, processing, transmission, provision, and disclosure of user personal information all comply with relevant laws and regulations and do not violate public order and good morals. The use of user data in this technical solution follows relevant national laws and regulations (e.g., the "Information Security Technology - Personal Information Security Specification"). For example: appropriate measures are taken for personal information access control; restrictions are imposed on the display of personal information; the purpose of using personal information does not exceed the scope of direct or reasonable association; and explicit identity targeting is eliminated when using personal information to avoid precisely locating a specific individual.

[0026] To address at least one of the technical problems existing in the aforementioned related technologies, the present invention provides a method and system for identifying defects in the structure of overhead contact line droppers; This embodiment discloses a defect identification method for abnormal overhead contact line dropper structures, such as... Figure 1 ,include: S100. Acquire images of overhead contact line components and extract the dropper area from the images of the overhead contact line components; in this embodiment, the specific method for extracting the dropper area from the images of overhead contact line components in S100 includes: S101. Collect a large number of images of catenary components containing droppers, and label the dropper tension cable clamps and dropper contact wire clamps in the images; S102. Train the target detection model, which is used to achieve precise positioning of the wire clamp; specifically, a deep learning target detection model is used to train the labeled data. The obtained model is able to detect dropper structures in images of various types of catenary components (4C).

[0027] S103. Obtain the complete dropper region using minimum bounding rectangle or KM weighted bipartite graph matching. In this embodiment, S103 uses minimum bounding rectangle or KM weighted bipartite graph matching to obtain the complete dropper region. The specific method includes: If the catenary tension cable clamp and catenary contact wire clamp are not detected in the image of the catenary components, then the catenary area is determined to be missing. If there is only one dropper cable clamp and one dropper contact wire clamp in the image of the overhead contact system components, the complete dropper area is obtained by using the minimum bounding rectangle method. If the image of the overhead contact system components contains more than one dropper cable clamp and one dropper contact wire clamp, the weighted bipartite graph minimum matching algorithm is used to match the pairs of dropper cable clamps and contact wire clamps.

[0028] S200. The suspension wire region is standardized using a Spatial Transformation Network (STN) to obtain a standardized image with a uniform posture; In this embodiment, the suspension wire region is standardized using a Spatial Transformation Network (STN). The specific method includes: S201. Using a localization network, generate an affine transformation matrix to describe the spatial transformation of the image of the drop wire region; S202. The mesh generator generates a corresponding mesh based on the affine transformation matrix, thereby realizing the spatial transformation of the input image; S203. The sampler uses bilinear interpolation to output a standardized image.

[0029] In this embodiment, the 4C image data is obtained by photographing the overhead contact line using a high-speed inspection vehicle. During this process, the dropper morphology in the 4C images will vary due to factors such as the vehicle's own characteristics (speed, vibration, tilt, etc.) and differences in the dropper installation caused by variations in the conductor pull-out value. This variation manifests primarily as different degrees of dropper tilt and inconsistent orientation of the dropper current-carrying rings. To ensure stability in subsequent feature analysis, the dropper images need to be preprocessed to improve their uniformity. The method used here is an STN network.

[0030] STN (Spatial Transformer) is a neural network architecture that adaptively transforms the spatial representation of input images, thereby improving the model's accuracy and robustness. An STN network consists of three main components: a Localization Network, a Grid Generator, and a Sampler. Currently, STN neural networks are used for standardization processing in the field of overhead contact line defect detection.

[0031] The localization network is a convolutional neural network whose output is an affine transformation matrix that describes the spatial transformation of the input image. The mesh generator applies the affine transformation matrix to the input image, generating a mesh that can be used to sample the input image. The sampler uses the generated mesh to sample the input image, thus producing the output image.

[0032] Specifically, the main function of the localization network is to learn an affine transformation matrix. Used to describe the spatial transformation of the input image. It is initialized as an identity transformation matrix at the beginning of network training and is continuously corrected through network iteration and loss function. Ultimately, transformations such as rotation, scaling, translation, and distortion are achieved.

[0033] The mesh generator applies an affine transformation matrix to the input image, generating a mesh that can be used to sample the input image. The output of the mesh generator is a two-dimensional mesh, where each point represents the location of a pixel in the output image. The mesh generator can generate a corresponding mesh based on the spatial transformation of the input image, thus achieving the spatial transformation of the input image. The formula for the mesh generator is as follows: ; here Representative input image The coordinates of the point Representative output image The most punctuation mark, among which and The coordinates are normalized to -1 to 1.

[0034] After the spatial transformation is completed, a sampler is needed to calculate the pixel values ​​corresponding to the transformed coordinate points. Here, a calculation based on bilinear interpolation is used: ; To output the first feature map The gray value of a certain point in a channel It is the first on the input feature map Channel points The grayscale. When or When it is greater than 1, the corresponding The value is 0, meaning only The grayscale values ​​of the surrounding four points determine the grayscale value of the target pixel. And when... and The smaller the value, the greater the impact and the greater the weight.

[0035] The diagrams of the dropper images before and after the STN network transformation in step S200 are shown below. Figure 2 and 3 As shown, Figure 2 and Figure 3 The images shown are of the dropper before and after the STN network transformation. It can be seen that the dropper images after STN network processing have similar appearance characteristics (similar dropper tilt direction, tilt degree, and current-carrying ring orientation, etc.).

[0036] S300. The standardized image input is fused with an unsupervised anomaly detection network SA-AnoNet using a spatial attention mechanism to extract structural features and calculate feature differences; in this embodiment, the unsupervised anomaly detection network SA-AnoNet is constructed based on an improved Ganomaly, and the specific construction method includes: Skip-Connection is introduced to form AnoNet, which solves the gradient vanishing problem in deep networks and improves reconstruction accuracy; An embedded spatial attention mechanism module is used to perform channel pooling, convolution, and sigmoid activation on the feature map to generate a spatial weight map and extract key structural features.

[0037] Specifically, after standardizing the dropper images, it is necessary to determine whether there are structural problems with the droppers. The traditional approach is to use a supervised classification model for identification, but this method has the following two drawbacks: a) A large number of defective samples are needed to build a classification model with generalizability, which is often not feasible in practice.

[0038] b) A model needs to be built for each specific defect type, making it impossible to identify unknown defect types.

[0039] To address the aforementioned drawbacks, we first introduced an anomaly detection network with a Skip-Connection structure, referred to as AnoNet.

[0040] AnoNet is a modification of the Ganomaly network. The Ganomaly network structure consists of three parts: an encoder and decoder forming the generator G, a discriminator D, and a reconstructed image feature extractor, sharing the same structure as the original. The input image is processed to obtain a high-dimensional feature vector, which is then processed to obtain the reconstructed image; features are extracted from the reconstructed image. The discriminator D determines the original input as true and the reconstructed image as false. This continuously optimizes the encoder and decoder in the generator, making the reconstructed image indistinguishable from the original image.

[0041] Directly using Ganomaly networks presents a significant problem: due to the complex and diverse acquisition environments of drop wire images, a relatively complex network (encoder GE(x) and decoder GD(z)) is required to improve the accuracy of feature extraction. This often leads to gradient vanishing, failure to converge, and ultimately, network unusability. To address this issue, we employ a Skip-Connection structure to connect the encoder GE(x) and decoder GD(z), resulting in AnoNet. The specific structure of AnoNet is as follows... Figure 4 As shown.

[0042] In this embodiment, the method for calculating feature differences in the unsupervised anomaly detection network SA-AnoNet includes: During the training phase, only normal dropper images are input to learn the standard structural feature distribution; During the inference phase, the distance between the standard structural features and the reconstructed structural features is calculated. The specific calculation formula includes: ; Where Diff(x) is the feature difference value, used to quantify the degree of difference between the input image and its reconstructed image in the feature space; GE(x) is the high-dimensional feature vector extracted by the encoder GE from the input image x; E(G(x)) is the high-dimensional feature extracted by the encoder GE after the input image x is reconstructed by encoding and decoding.

[0043] Specifically, for image segmentation tasks or tasks focusing on spatial domain information, the pooling layers in traditional network encoders significantly reduce the size of feature maps, which is detrimental to image reconstruction. Skip-Connection allows the network to fuse feature maps at corresponding positions in the encoder across channels during upsampling at each level. By fusing low-level and high-level features, the network can retain more high-resolution detail information contained in high-level feature maps, thereby improving reconstruction accuracy. In-neighbor layer feature reconstruction and cross-layer feature reconstruction enhance the transfer of feature information between layers while further utilizing the rich detail information in high-level convolutional feature layers, thus maximizing the utilization rate of feature information in each layer of the network.

[0044] The AnoNet network is trained using standardized dropper images without structural anomalies. When a standardized dropper image with structural anomalies is input, the encoder and decoder in the model become inapplicable to the anomaly. The decoder adds features only present in normal samples to the reconstructed image of the anomaly. This results in a significant difference between the encoded high-dimensional feature vector and the actual image. The localization difference value is: ; Set threshold ,when If so, the input image can be identified as an abnormal image.

[0045] Secondly, in order to further enable the AnoNet network to focus on feature extraction of the dropper structure information and reduce the impact of various objective conditions such as imaging background, lighting changes, vehicle movement, and lens blur on the feature extraction of the dropper image, we further introduce a spatial attention mechanism into the network structure to improve the robustness and stability of AnoNet feature extraction.

[0046] In the AnoNet encoder GE(x) of this implementation, for each feature map F' of size C×H×W, max pooling and average pooling are used in the channel dimension to obtain two two-dimensional features, which are then concatenated together in the channel dimension, resulting in 2×H×W features. A 7×7 convolution and a sigmoid activation function are then used to obtain the weight coefficients Ms. Finally, Ms×F' yields the scaled new features. The spatial attention mechanism effectively alters the response of AnoNet network features to spatial location, increasing the feature weights of important regions.

[0047] S400. The feature difference is compared with a preset threshold to identify defects in the dropper structure. In this embodiment, the method for identifying defects in the dropper structure by comparing the feature difference with a preset threshold includes: when the feature difference Diff(x) > the preset threshold... When it is determined that there is an abnormal defect in the dropper structure, the abnormal defect includes at least the missing dropper current-carrying ring, the dropper not being under force, the dropper being missing, and the dropper current-carrying ring being installed in the wrong direction.

[0048] This embodiment provides a defect identification method for catenary dropper structures, comprising: acquiring images of catenary components and extracting dropper regions from the images; standardizing the dropper regions using a Spatial Transformation Network (STN) to obtain standardized images with uniform posture; inputting the standardized images into an unsupervised anomaly detection network (SA-AnoNet) incorporating a spatial attention mechanism to extract structural features and calculate feature differences; and comparing the feature differences with a preset threshold to identify defects in the dropper structure. This invention requires no defect samples for training, exhibits strong anti-interference capabilities, high standardization, and fully automated operation. It can quickly identify defects such as current-carrying loop anomalies, lack of force, missing loops, offsets, and height exceeding limits, significantly improving detection efficiency and accuracy. It is suitable for intelligent operation and maintenance of catenary droppers in electrified railways.

[0049] Based on the same inventive concept, this embodiment discloses a defect identification system for abnormal overhead contact line dropper structures, such as... Figure 5 It includes: a dropper area extraction module, a standardization module, an anomaly detection module, and a result output module; among which: The dropper area extraction module is used to acquire images of catenary components and extract the dropper area from the images of the catenary components. The standardization module is used to standardize the dropper region through a spatial transformation network (STN) to obtain a standardized image with a uniform posture. The anomaly detection module is used to integrate standardized image inputs with the unsupervised anomaly detection network SA-AnoNet, which uses a spatial attention mechanism, to extract structural features and calculate feature differences. The result output module is used to compare the feature differences with a preset threshold to identify defects in the suspension wire structure.

[0050] The specific working methods of the dropper area extraction module, standardization module, anomaly identification module, and result output module have been described in detail in a defect identification method for anomalies in the contact wire dropper structure, and will not be repeated here.

[0051] Based on the same inventive concept, embodiments of the present invention also provide an electronic device. Figure 6 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Figure 6As shown, an embodiment of the present invention provides an electronic device including: one or more processors 101, a memory 102, and one or more I / O interfaces 103. The memory 102 stores one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement any of the identification methods described in the above embodiments; the one or more I / O interfaces 103 are connected between the processor and the memory, configured to enable information interaction between the processor and the memory.

[0052] The processor 101 is a device with data processing capabilities, including but not limited to a central processing unit (CPU); the memory 102 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and flash memory (FLASH); the I / O interface (read / write interface) 103 is connected between the processor 101 and the memory 102, and can realize information interaction between the processor 101 and the memory 102, including but not limited to a data bus (Bus).

[0053] In some embodiments, the processor 101, memory 102, and I / O interface 103 are interconnected via bus 104, and thus connected to other components of the computing device.

[0054] In some embodiments, the one or more processors 101 include a field-programmable gate array.

[0055] This invention also provides a computer-readable medium. The computer-readable medium stores a computer program, which, when executed by a processor, implements the steps of any of the identification methods described in the above embodiments. The computer-readable storage medium may be volatile or non-volatile.

[0056] This invention also provides a computer program product, including computer-readable code, or a non-volatile computer-readable storage medium carrying computer-readable code. When the computer-readable code is run in the processor of an electronic device, the processor in the electronic device executes the above-described identification method.

[0057] Those skilled in the art will understand that all or some of the steps, systems, and apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software can be distributed on a computer-readable storage medium, which may include computer storage media (or non-transitory media) and communication media (or transient media).

[0058] As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable program instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technologies, portable compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable program instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0059] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0060] The computer program instructions used to perform the operations of this invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the computer-readable program instructions. This electronic circuitry can execute the computer-readable program instructions to implement various aspects of the invention.

[0061] The computer program product described herein can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0062] Various aspects of the present invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0063] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0064] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0065] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0066] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for illustrative purposes only and should be construed as such, and is not intended to be limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in conjunction with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in conjunction with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of the invention as set forth in the appended claims.

Claims

1. A method for identifying defects in the structure of overhead contact line droppers, characterized in that, include: Acquire images of overhead contact system components and extract the dropper area from the images of the overhead contact system components; The suspension wire region is standardized by using a Spatial Transformation Network (STN) to obtain a standardized image with a uniform posture. The unsupervised anomaly detection network SA-AnoNet, which integrates standardized image input with spatial attention mechanism, extracts structural features and calculates feature differences. The defects of abnormal suspension wire structure are identified by comparing the feature differences with a preset threshold.

2. The method according to claim 1, characterized in that, Extracting the dropper area from images of overhead contact system components, specifically including: Collect a large number of images of catenary components containing droppers, and label the dropper tension cable clamps and dropper contact wire clamps in the images; Train a target detection model, which is used to achieve precise positioning of the wire clamp; The complete dropper region is obtained by using the minimum bounding rectangle or KM weighted bipartite graph matching.

3. The method according to claim 2, characterized in that, The complete dropper region can be obtained by using minimum bounding rectangle or KM weighted bipartite graph matching. Specific methods include: If the catenary tension cable clamp and catenary contact wire clamp are not detected in the image of the catenary components, then the catenary area is determined to be missing. If there is only one dropper cable clamp and one dropper contact wire clamp in the image of the overhead contact system components, the complete dropper area is obtained by using the minimum bounding rectangle method. If the image of the overhead contact system components contains more than one dropper cable clamp and one dropper contact wire clamp, the weighted bipartite graph minimum matching algorithm is used to match the pairs of dropper cable clamps and contact wire clamps.

4. The method according to claim 1, characterized in that, The suspension string region is standardized using a Spatial Transformation Network (STN). Specific methods include: A localization network is used to generate an affine transformation matrix to describe the spatial transformation of the image of the drop wire region; The mesh generator generates a corresponding mesh based on the affine transformation matrix, thereby realizing the spatial transformation of the input image. The sampler uses bilinear interpolation to output a standardized image.

5. The method according to claim 4, characterized in that, The formula for the mesh generator is as follows: ； here Representative input image The coordinates of the point Representative output image The most punctuation mark, Represents the affine transformation matrix, where and The coordinates are normalized to -1 to 1.

6. The method according to claim 1, characterized in that, The unsupervised anomaly detection network SA-AnoNet is built based on an improved Ganomaly, and the specific construction method includes: Skip-Connection is introduced to form AnoNet, which solves the gradient vanishing problem in deep networks and improves reconstruction accuracy; An embedded spatial attention mechanism module is used to perform channel pooling, convolution, and sigmoid activation on the feature map to generate a spatial weight map and extract key structural features.

7. The method according to claim 6, characterized in that, The method for calculating feature differences in the unsupervised anomaly detection network SA-AnoNet includes: During the training phase, only normal dropper images are input to learn the standard structural feature distribution; During the inference phase, the distance between the standard structural features and the reconstructed structural features is calculated. The specific calculation formula includes: ； Where Diff(x) is the feature difference value, used to quantify the degree of difference between the input image and its reconstructed image in the feature space; GE(x) is the high-dimensional feature vector extracted by the encoder GE from the input image x; E(G(x)) is the high-dimensional feature vector extracted by the encoder GE after the input image x is reconstructed by encoding and decoding.

8. The method according to claim 1, characterized in that, The defects in the suspension cable structure are identified by comparing the aforementioned feature differences with a preset threshold. Specific methods include: When the feature difference Diff(x) > preset threshold When it is determined that there is an abnormal defect in the dropper structure, the abnormal defect includes at least the missing dropper current-carrying ring, the dropper not being under force, the dropper being missing, and the dropper current-carrying ring being installed in the wrong direction.

9. A defect identification system for abnormal overhead contact line dropper structure, employing the method described in any one of claims 1-8, characterized in that, include: The system includes a dropper area extraction module, a standardization module, an anomaly detection module, and a result output module; among which: The dropper area extraction module is used to acquire images of catenary components and extract the dropper area from the images of the catenary components. The standardization module is used to standardize the dropper region through a spatial transformation network (STN) to obtain a standardized image with a uniform posture. The anomaly detection module is used to integrate standardized image inputs with the unsupervised anomaly detection network SA-AnoNet, which uses a spatial attention mechanism, to extract structural features and calculate feature differences. The result output module is used to compare the feature differences with a preset threshold to identify defects in the suspension wire structure.

10. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the identification method as described in any one of claims 1 to 8.

Images