Method and system for diagnosing faults in a cable and accessories thereof

Abstract

The application provides a cable and accessory fault diagnosis method and system, comprising: acquiring image information of a to-be-diagnosed area, inputting the image information of the to-be-diagnosed area into a preset neural network model for format conversion processing to obtain an image of a prediction area; superimposing and fusing the image information of the to-be-diagnosed area and the image of the prediction area to obtain a target image; randomly framing an image in the target image, calculating a corresponding temperature gradient value according to a corresponding temperature value in the framed image, and determining whether a fault exists in the framed image according to the temperature gradient value; and if a fault exists, determining that a cable and accessory in the area have a fault. The application effectively improves the recognition accuracy of an infrared image of a distribution network cable and accessory, realizes an automatic integrated fault diagnosis platform, and provides great convenience for maintenance personnel to regularly maintain the cable and accessory.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault diagnosis of cables and accessories thereof, and particularly relates to a fault diagnosis method and system of cables and accessories thereof. BACKGROUND

[0002] At present, the fault diagnosis of cables and accessories thereof under complex conditions puts great test on the professionalism and work experience of maintenance personnel and has poor diagnosis effect. With the rapid development of artificial intelligence technology, the diagnosis of power equipment is developing towards automation, and a fine fault diagnosis system is formed, and an automatic integrated fault diagnosis platform is established, which provides great convenience for maintenance personnel to regularly maintain cables and accessories.

[0003] At present, the environment of cables and accessories thereof is complex, so there are problems of difficult effective identification of equipment area and low diagnosis efficiency. SUMMARY

[0004] The present application aims to provide a fault diagnosis method and system of cables and accessories thereof, which solves the technical problem of difficult effective identification of equipment area and low diagnosis efficiency of cables and accessories thereof under complex environment.

[0005] In one aspect, a fault diagnosis method of cables and accessories thereof is provided, comprising:

[0006] obtaining image information of a to-be-diagnosed area and inputting the image information of the to-be-diagnosed area into a preset neural network model for format conversion processing to obtain an image of a predicted area;

[0007] superimposing and fusing the image information of the to-be-diagnosed area and the image of the corresponding predicted area to obtain a target image;

[0008] randomly framing an image in the target image, calculating a corresponding temperature gradient value according to a corresponding temperature value in the framed image, and determining whether there is a fault in the framed image according to the temperature gradient value, if there is a fault, determining that the cable and accessories in the area have a fault.

[0009] Preferably, the image of the predicted area is obtained, specifically comprising:

[0010] performing size modification on the obtained image information of the to-be-diagnosed area to obtain an infrared image conforming to a preset size;

[0011] identifying corresponding areas in the image information of the to-be-diagnosed area and the infrared image conforming to the preset size by a preset neural network model, extracting features by convolution pooling, restoring features by deconvolution for result prediction, and outputting an image of a predicted area conforming to the preset size.

[0012] Preferably, the process of overlaying and fusing the image information of the region to be diagnosed with the image of the corresponding predicted region specifically includes:

[0013] Filter the pixels in the image information of the area to be diagnosed according to the preset pixel filtering conditions, and determine the pixels of the cable or cable accessories in the image information of the area to be diagnosed.

[0014] Pixels within the image of the prediction region are filtered according to preset pixel filtering conditions to determine the pixels of the cable or cable accessory within the image of the prediction region.

[0015] The pixels of the cable or cable accessory within the image of the predicted region are superimposed onto the pixels of the cable or cable accessory within the image information of the region to be diagnosed, to obtain the target image.

[0016] Preferably, determining the pixels of the cable or cable accessory within the image information of the area to be diagnosed specifically includes:

[0017] Detect pixels within the image of the region to be diagnosed based on preset pixel filtering conditions;

[0018] If a pixel meets the preset pixel filtering criteria, it is determined that the pixel is a cable or cable accessory pixel and the pixel is retained.

[0019] If a pixel does not meet the preset pixel filtering criteria, it is determined that the pixel is not a pixel of a cable or cable accessory, and the pixel is deleted.

[0020] Preferably, determining the pixels of the cable or cable accessory within the image information of the predicted region specifically includes:

[0021] Detect pixels within the image of the predicted region according to preset pixel filtering conditions;

[0022] If a pixel meets the preset pixel filtering criteria, it is determined that the pixel is a cable or cable accessory pixel and the pixel is retained.

[0023] If a pixel does not meet the preset pixel filtering criteria, it is determined that the pixel is not a pixel of a cable or cable accessory, and the pixel is deleted.

[0024] Preferably, the step of calculating the corresponding temperature gradient value based on the temperature value within the selected image specifically includes:

[0025] In the target image, an image region is selected through a preset rectangle, and the coordinates of the upper left corner of the rectangle are obtained;

[0026] Identify the temperature value represented by the infrared image within the rectangle, and calculate the corresponding temperature gradient value based on the identified temperature value and the coordinates of the upper left corner of the rectangle.

[0027] Preferably, the corresponding temperature gradient value is calculated using the following formula:

[0028]

[0029] in, Let T represent the gradient at the point to be determined. a Let x represent the temperature of the point to be determined, and xy represent the coordinates of the top-left corner of the rectangle. This represents the undirected gradient value corresponding to x. This represents the directed vector value corresponding to x. This represents the undirected gradient value corresponding to x. This represents the directional vector value corresponding to x.

[0030] Preferably, determining whether a fault exists within the selected image based on the temperature gradient value specifically includes:

[0031] The temperature gradient value corresponding to the selected image is compared with the preset judgment criteria. If the temperature gradient value corresponding to the selected image meets the preset judgment criteria, the cable or cable accessory corresponding to the selected image is determined to be fault-free. If the temperature gradient value corresponding to the selected image does not meet the preset judgment criteria, the cable or cable accessory corresponding to the selected image is determined to be faulty.

[0032] On the other hand, a fault diagnosis system for cables and their accessories is also provided, for implementing the fault diagnosis method for the cables and their accessories, including:

[0033] The image preprocessing module is used to acquire image information of the region to be diagnosed, and input the image information of the region to be diagnosed into a preset neural network model for format conversion processing to obtain the image of the predicted region;

[0034] The image overlay module is used to overlay and fuse the image information of the region to be diagnosed with the image of the corresponding prediction region to obtain the target image;

[0035] The fault determination module is used to randomly select an image within the target image, calculate the corresponding temperature gradient value based on the corresponding temperature value within the selected image, and determine whether there is a fault within the selected image based on the temperature gradient value. If a fault exists, it is determined that the cable and its accessories in that area are faulty.

[0036] Preferably, the fault judgment module is further used to compare the temperature gradient value corresponding to the selected image with a preset judgment standard. If the temperature gradient value corresponding to the selected image meets the preset judgment standard, it is determined that the cable or cable accessory corresponding to the selected image is fault-free. If the temperature gradient value corresponding to the selected image does not meet the preset judgment standard, it is determined that the cable or cable accessory corresponding to the selected image is faulty.

[0037] In summary, implementing the embodiments of the present invention has the following beneficial effects:

[0038] The fault diagnosis method and system for cables and their accessories provided by this invention firstly achieves regional identification and key temperature information acquisition in cable trenches and tunnels by designing a fully convolutional neural network model. Then, by introducing the concept of temperature gradient, a coordinated diagnostic model is designed to achieve early cable fault diagnosis in cable trenches and tunnels. This effectively improves the recognition accuracy of infrared images of distribution network cables and their accessories, realizing an automated and integrated fault diagnosis platform, and providing significant convenience for maintenance personnel to regularly inspect and maintain cables and their accessories. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, obtaining other drawings based on these drawings without creative effort still falls within the scope of the present invention.

[0040] Figure 1 This is a schematic diagram of the main flow of a fault diagnosis method for cables and their accessories according to an embodiment of the present invention.

[0041] Figure 2 This is a schematic diagram of a fault diagnosis system for cables and their accessories according to an embodiment of the present invention.

[0042] Figure 3 This is a schematic diagram of a neural network model in an embodiment of the present invention.

[0043] Figure 4 This is a schematic diagram of a neural network model in an embodiment of the present invention.

[0044] Figure 5 This is a schematic diagram of image overlay and fusion in an embodiment of the present invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings.

[0046] like Figure 1The diagram shown is a schematic representation of an embodiment of a fault diagnosis method for cables and their accessories provided by the present invention. In this embodiment, the method includes the following steps:

[0047] Image information of the region to be diagnosed is acquired and input into a preset neural network model for format conversion processing to obtain the image of the predicted region. In this embodiment, acquiring the image information of the region to be diagnosed refers to acquiring an infrared image. Infrared images captured by existing commercially available devices are typically (640, 480, 3). Therefore, in order to input the image into the model for prediction, the original infrared image is first resized to (480, 480, 3).

[0048] In this embodiment, the image information of the region to be diagnosed is resized to obtain an infrared image that conforms to a preset size. A preset neural network model is used to identify the corresponding regions in the image information of the region to be diagnosed and the infrared image conforming to the preset size. Features are extracted through convolutional pooling and restored through deconvolution to predict the results, outputting an image of the predicted region conforming to the preset size. Essentially, the resizing principle uses nonlinear interpolation to establish the correspondence between two regions. The resized image is then input into the built model, where features are extracted through convolutional pooling and restored through deconvolution to predict the results. Finally, an image of (480, 480, 3) is output, which is the image of the corresponding predicted region. The built model refers to a fully convolutional neural network model (a preset neural network model), such as... Figure 3 As shown. It should be noted that using a neural network model for the above preprocessing is a common practice, and will not be elaborated further here.

[0049] Furthermore, the image information of the region to be diagnosed is superimposed and fused with the image of the corresponding prediction region to obtain the target image; that is, an image superimposition and fusion method based on OpenCV uses the idea of ​​image processing to superimpose and fuse the prediction image of the fully convolutional neural network with the original input image as the standard input image of the fault diagnosis model.

[0050] In this embodiment, pixels within the image information of the region to be diagnosed are filtered according to preset pixel filtering conditions to determine pixels representing cables or cable accessories within the image information of the region to be diagnosed. Specifically, pixels within the image of the region to be diagnosed are detected according to preset pixel filtering conditions. If a pixel meets the preset pixel filtering conditions, it is determined that the pixel is a cable or cable accessory pixel and is retained. If a pixel does not meet the preset pixel filtering conditions, it is determined that the pixel is not a cable or cable accessory pixel and is deleted. Similarly, pixels within the image of the prediction region are filtered according to preset pixel filtering conditions to determine pixels representing cables or cable accessories within the image of the prediction region. Specifically, pixels within the image of the prediction region are detected according to preset pixel filtering conditions. If a pixel meets the preset pixel filtering conditions, it is determined that the pixel is a cable or cable accessory pixel and is retained. If a pixel does not meet the preset pixel filtering conditions, it is determined that the pixel is not a cable or cable accessory pixel and is deleted. Understandably, for the input and output of the fully wound neural network model, both are (480, 480, 3), meaning the pixels correspond to each other. Using OR and NOT mathematical logic, pixels in the predicted image output by the fully wound neural network model that meet the condition (i.e., the target is a cable or cable accessory) retain the values ​​of the corresponding pixels in the original input image. Pixels that do not meet the above conditions are deleted from the original input image. Finally, the superimposed and fused images become the input image for the fault diagnosis model. Figure 5 As shown.

[0051] Specifically, the pixels of the cable or cable accessory in the image of the predicted region are superimposed onto the pixels of the cable or cable accessory in the image information of the region to be diagnosed to obtain the target image.

[0052] Furthermore, by randomly selecting an image within the target image, a corresponding temperature gradient value is calculated based on the temperature value within the selected image. This temperature gradient value is then used to determine whether a fault exists within the selected image. If a fault is found, the cable and its accessories within that area are determined to be faulty. In other words, this is based on the idea of ​​transfer learning, where transfer refers to using MobileNet as a feature extraction method in a convolutional network model, and finally extracting the classification results through fully connected and Softmax layers. This invention, based on the trained model, has achieved effective fault diagnosis of cables and their accessories. Its model block diagram is shown below. Figure 4 As shown.

[0053] In this embodiment, an image region is selected from the target image using a preset rectangular frame, and the coordinates of the upper left corner of the rectangle are obtained. The temperature value represented by the infrared image within the rectangle is identified, and the corresponding temperature gradient value is calculated based on the identified temperature value and the coordinates of the upper left corner of the rectangle. The corresponding temperature gradient value is calculated using the following formula:

[0054]

[0055] in, Let T represent the gradient at the point to be determined. a Let x represent the temperature of the point to be determined, and xy represent the coordinates of the top-left corner of the rectangle. This represents the undirected gradient value corresponding to x. This represents the directed vector value corresponding to x. This represents the undirected gradient value corresponding to x. This represents the directional vector value corresponding to x.

[0056] Understandably, corresponding images have corresponding annotation files. Since the model's purpose is fault diagnosis of cables and their accessories, the annotation primarily involves selecting the fault area using a rectangular bounding box. The corresponding annotation file content is (fault, x, y, height, width), where the first value represents the presence or absence of a fault, the second and third represent the coordinates of the top-left corner of the rectangle, and the fourth and fifth represent the height and width of the rectangle. To address the issue of gradients characterizing the maximum change in a region, the concept of a temperature gradient is introduced. During the temperature gradient calculation process... and It is a value without direction. and Directed vector values. After solving the above parts, the results are input into the model for fault diagnosis. Finally, compared with the traditional CNN recognition network MobileNet, which uses color infrared images as input, this model uses color infrared images as input and is trained using color-annotated images. The fault diagnosis method for distribution network cables and their accessories based on collaborative deep learning designed in this invention has high recognition accuracy and significant defect diagnosis effect, and the recognition and classification effect is more detailed than that of traditional CNN classification.

[0057] Specifically, the temperature gradient value corresponding to the selected image is compared with a preset judgment standard. If the temperature gradient value corresponding to the selected image meets the preset judgment standard, the cable or cable accessory corresponding to the selected image is determined to be fault-free. If the temperature gradient value corresponding to the selected image does not meet the preset judgment standard, the cable or cable accessory corresponding to the selected image is determined to be faulty.

[0058] like Figure 2As shown, the present invention also provides a fault diagnosis system for cables and their accessories, and a fault diagnosis method for the aforementioned cables and their accessories, comprising:

[0059] The image preprocessing module is used to acquire image information of the region to be diagnosed, and input the image information of the region to be diagnosed into a preset neural network model for format conversion processing to obtain the image of the predicted region;

[0060] The image overlay module is used to overlay and fuse the image information of the region to be diagnosed with the image of the corresponding prediction region to obtain the target image;

[0061] The fault determination module is used to randomly select an image within the target image, calculate the corresponding temperature gradient value based on the corresponding temperature value within the selected image, and determine whether there is a fault within the selected image based on the temperature gradient value. If a fault exists, it is determined that the cable and its accessories in that area are faulty.

[0062] In this embodiment, the fault judgment module is further used to compare the temperature gradient value corresponding to the selected image with a preset judgment standard. If the temperature gradient value corresponding to the selected image meets the preset judgment standard, it is determined that the cable or cable accessory corresponding to the selected image is fault-free. If the temperature gradient value corresponding to the selected image does not meet the preset judgment standard, it is determined that the cable or cable accessory corresponding to the selected image is faulty.

[0063] It should be noted that the system described in the above embodiments corresponds to the method described in the above embodiments. Therefore, the parts of the system described in the above embodiments that are not described in detail can be obtained by referring to the content of the method described in the above embodiments, and will not be repeated here.

[0064] In summary, implementing the embodiments of the present invention has the following beneficial effects:

[0065] The fault diagnosis method and system for cables and their accessories provided by this invention firstly achieves regional identification and key temperature information acquisition in cable trenches and tunnels by designing a fully convolutional neural network model. Then, by introducing the concept of temperature gradient, a coordinated diagnostic model is designed to achieve early cable fault diagnosis in cable trenches and tunnels. This effectively improves the recognition accuracy of infrared images of distribution network cables and their accessories, realizing an automated and integrated fault diagnosis platform, and providing significant convenience for maintenance personnel to regularly inspect and maintain cables and their accessories.

[0066] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for fault diagnosis of cables and their accessories, characterized in that, include: Image information of the region to be diagnosed is acquired, and the image information of the region to be diagnosed is input into a preset neural network model for format conversion processing to obtain the image of the predicted region; The image information of the region to be diagnosed is superimposed and fused with the image of the corresponding predicted region to obtain the target image; By randomly selecting an image within the target image, the corresponding temperature gradient value is calculated based on the temperature value within the selected image, and the presence of a fault within the selected image is determined based on the temperature gradient value. If a fault exists, the cable and its accessories within that area are determined to be faulty. Specifically, obtaining the image of the predicted region includes: The image information of the area to be diagnosed is resized to obtain an infrared image that conforms to a preset size; The system uses a pre-defined neural network model to identify the image information of the region to be diagnosed and the corresponding region in an infrared image of a pre-defined size. After extracting features through convolutional pooling and restoring features through deconvolution, the system predicts the result and outputs an image of the predicted region that conforms to the pre-defined size. The preset neural network model takes into account the influence of intermediate features during feature extraction, and fuses feature layers of the same size in the convolution and deconvolution processes; The process of overlaying and fusing the image information of the region to be diagnosed with the image of the corresponding predicted region specifically includes: Filter the pixels in the image information of the area to be diagnosed according to the preset pixel filtering conditions, and determine the pixels of the cable or cable accessories in the image information of the area to be diagnosed. Pixels within the image of the prediction region are filtered according to preset pixel filtering conditions to determine the pixels of the cable or cable accessory within the image of the prediction region. The pixels of the cable or cable accessory in the image of the predicted region are superimposed onto the pixels of the cable or cable accessory in the image information of the region to be diagnosed to obtain the target image; The step of calculating the corresponding temperature gradient value based on the temperature value within the selected image specifically includes: In the target image, an image region is selected through a preset rectangle, and the coordinates of the upper left corner of the rectangle are obtained; Identify the temperature value represented by the infrared image within the rectangle, and calculate the corresponding temperature gradient value based on the identified temperature value and the coordinates of the top-left corner of the rectangle; and calculate the corresponding temperature gradient value using the following formula: in, This represents the gradient at the point to be determined. Let x represent the temperature of the point to be determined, and xy represent the coordinates of the top-left corner of the rectangle. This represents the undirected gradient value corresponding to x. This represents the directed vector value corresponding to x. This represents the undirected gradient value corresponding to x. This represents the directed vector value corresponding to x; The corresponding image has a corresponding annotation file, the content of which is (fault, x, y, height, width). The first value represents whether there is a fault, the second and third values ​​represent the coordinates of the upper left corner of the rectangle, and the fourth and fifth values ​​represent the height and width of the rectangle. After training the fault diagnosis model using this annotation file, the calculated temperature gradient value is input into the fault diagnosis model for fault diagnosis.

2. The method as described in claim 1, characterized in that, The step of determining the pixels of the cable or cable accessory within the image information of the region to be diagnosed specifically includes: Detect pixels within the image of the region to be diagnosed based on preset pixel filtering conditions; If a pixel meets the preset pixel filtering criteria, it is determined that the pixel is a cable or cable accessory pixel and the pixel is retained. If a pixel does not meet the preset pixel filtering criteria, it is determined that the pixel is not a pixel of a cable or cable accessory, and the pixel is deleted.

3. The method as described in claim 1, characterized in that, The determination of the pixels of the cable or cable accessory within the image information of the predicted region specifically includes: Detect pixels within the image of the predicted region according to preset pixel filtering conditions; If a pixel meets the preset pixel filtering criteria, it is determined that the pixel is a cable or cable accessory pixel and the pixel is retained. If a pixel does not meet the preset pixel filtering criteria, it is determined that the pixel is not a pixel of a cable or cable accessory, and the pixel is deleted.

4. The method as described in claim 2 or 3, characterized in that, The step of determining whether a fault exists within the selected image based on the temperature gradient value specifically includes: The temperature gradient value corresponding to the selected image is compared with the preset judgment criteria. If the temperature gradient value corresponding to the selected image meets the preset judgment criteria, the cable or cable accessory corresponding to the selected image is determined to be fault-free. If the temperature gradient value corresponding to the selected image does not meet the preset judgment criteria, the cable or cable accessory corresponding to the selected image is determined to be faulty.

5. A fault diagnosis system for cables and their accessories, for implementing the method as described in any one of claims 1-4, characterized in that, include: The image preprocessing module is used to acquire image information of the region to be diagnosed, and input the image information of the region to be diagnosed into a preset neural network model for format conversion processing to obtain the image of the predicted region; The image overlay module is used to overlay and fuse the image information of the region to be diagnosed with the image of the corresponding prediction region to obtain the target image; The fault determination module is used to randomly select an image within the target image, calculate the corresponding temperature gradient value based on the corresponding temperature value within the selected image, and determine whether there is a fault within the selected image based on the temperature gradient value. If a fault exists, it is determined that the cable and its accessories in that area are faulty.

6. The system as described in claim 5, characterized in that, The fault judgment module is also used to compare the temperature gradient value corresponding to the selected image with a preset judgment standard. If the temperature gradient value corresponding to the selected image meets the preset judgment standard, it is determined that the cable or cable accessory corresponding to the selected image is fault-free. If the temperature gradient value corresponding to the selected image does not meet the preset judgment standard, it is determined that the cable or cable accessory corresponding to the selected image is faulty.

Images