Fault detection method and device of power transformation equipment rubber sleeve and storage medium

Abstract

The invention discloses a fault detection method and device for a power transformation equipment rubber sleeve and a storage medium. The method comprises the steps that at least one target rubber sleeve image of a target power transformation equipment rubber sleeve is acquired; inputting the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, and determining a rubber sleeve fault detection result based on a model output result; wherein the rubber sleeve fault detection model is obtained by training a pre-established initial convolutional neural network based on a sample rubber sleeve image, and the rubber sleeve fault detection result comprises fault category information, confidence corresponding to the fault category information and fault marking box position information; and generating rubber sleeve fault prompt information corresponding to the target power transformation equipment according to the rubber sleeve fault detection result, and displaying the rubber sleeve fault prompt information. And the fault detection efficiency and accuracy of the power transformation equipment rubber sleeve are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault detection of power substation equipment, and in particular to a fault detection method, device and storage medium for a rubber sleeve of power substation equipment. Background Art

[0002] With the development of intelligent and automated power systems, substation equipment, as a core component of the power system, has become particularly critical for health monitoring of its operating status. Rubber sleeve failure in substation equipment may cause serious insulation problems, resulting in equipment damage or even power outages. Therefore, accurate and real-time detection of substation equipment rubber sleeve failure is crucial for the safe operation of the power system.

[0003] At present, most traditional rubber sleeve fault detection methods rely on manual inspection or visual inspection methods. The efficiency of rubber sleeve fault detection is low, the labor cost is high, and it is difficult to perform efficient status monitoring in large-scale power equipment. In complex environments, manual inspections are prone to miss minor faults and fail to detect early rubber sleeve fault problems in time. In addition, the visual inspection method has high requirements on the quality of the acquired rubber sleeve image. In actual scenarios, environmental noise, lighting changes, and equipment structure complexity often lead to low detection accuracy. Summary of the invention

[0004] The present invention provides a fault detection method, device and storage medium for a rubber sleeve of a power transformation equipment, so as to solve the problem of low efficiency and low accuracy in rubber sleeve fault detection.

[0005] According to one aspect of the present invention, a fault detection method for a rubber sleeve of a substation equipment is provided, the method comprising: obtaining at least one target rubber sleeve image of a target substation equipment rubber sleeve; inputting the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, and determining a rubber sleeve fault detection result based on the model output result; wherein the rubber sleeve fault detection model is obtained by training a pre-established initial convolutional neural network based on sample rubber sleeve images, and the rubber sleeve fault detection result comprises fault category information, a confidence level corresponding to the fault category information, and fault marking box position information; generating rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result, and displaying the rubber sleeve fault prompt information.

[0006] According to another aspect of the present invention, a fault detection device for a rubber sleeve of a substation equipment is provided, and the device includes: an image acquisition module, used to acquire at least one target rubber sleeve image of a target substation equipment rubber sleeve; a detection result determination module, used to input the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, and determine the rubber sleeve fault detection result based on the model output result; wherein the rubber sleeve fault detection model is obtained by training a pre-established initial convolutional neural network based on a sample rubber sleeve image, and the rubber sleeve fault detection result includes fault category information, a confidence level corresponding to the fault category information, and a fault marking box position information; a prompt module, used to generate rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result, and display the rubber sleeve fault prompt information.

[0007] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0008] at least one processor; and

[0009] a memory communicatively connected to the at least one processor; wherein,

[0010] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the fault detection method for the rubber sleeve of the substation equipment described in any embodiment of the present invention.

[0011] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the fault detection method for a rubber sleeve of a substation equipment described in any embodiment of the present invention when executed.

[0012] The technical solution of the embodiment of the present invention is to obtain at least one target rubber sleeve image of the rubber sleeve of the target substation equipment; obtain the image of the rubber sleeve of the target substation equipment so as to perform fault detection based on the target rubber sleeve image later; then, input the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, and determine the rubber sleeve fault detection result based on the model output result; wherein, the rubber sleeve fault detection model is obtained by training a pre-established initial convolutional neural network based on the sample rubber sleeve image, and the rubber sleeve fault detection result includes fault category information, confidence corresponding to the fault category information, and fault marking box position information; perform fault detection on the target rubber sleeve image through the rubber sleeve fault detection model to improve detection efficiency and accuracy; finally, generate rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result, and display the rubber sleeve fault prompt information. The problem of low efficiency and low accuracy of rubber sleeve fault detection is solved, and the beneficial effect of improving the fault detection efficiency and accuracy of the rubber sleeve of the substation equipment is achieved.

[0013] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0015] Figure 1 This is a flow chart of a fault detection method for a rubber sleeve of a substation equipment provided in accordance with the first embodiment of the present invention;

[0016] Figure 2 This is a flow chart of a fault detection method for a rubber sleeve of a substation equipment provided in accordance with a second embodiment of the present invention;

[0017] Figure 3 It is a structural schematic diagram of a fault detection device for a rubber sleeve of a substation equipment provided according to a third embodiment of the present invention;

[0018] Figure 4 The present invention is a schematic diagram of the structure of an electronic device for implementing the fault detection method for a rubber sleeve of a power transformation equipment according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0020] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0021] Embodiment 1

[0022] Figure 1 A flowchart of a method for detecting a fault of a rubber sleeve of a transformer substation is provided for the first embodiment of the present invention. This embodiment is applicable to detecting a fault of a rubber sleeve of a transformer substation. The method can be performed by a fault detection device for a rubber sleeve of a transformer substation. The fault detection device for a rubber sleeve of a transformer substation can be implemented in the form of hardware and / or software. The fault detection device for a rubber sleeve of a transformer substation can be configured in an electronic device. Figure 1 As shown, the method includes:

[0023] S110, obtaining at least one target rubber sleeve image of a target substation equipment rubber sleeve.

[0024] The target substation equipment rubber sleeve can be understood as the rubber sleeve of the substation equipment to be detected. The target rubber sleeve image can be understood as the rubber sleeve image to be detected.

[0025] Specifically, a high-definition camera is installed at a key position of the substation to ensure that the field of view can fully cover the substation equipment. The automatic shooting interval and focal length of the camera are set, and the target rubber sleeve image of the rubber sleeve of the target substation equipment is collected based on the automatic shooting interval and focal length.

[0026] S120, inputting the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, and determining a rubber sleeve fault detection result based on a model output result.

[0027] The rubber sleeve fault detection model is obtained by training a pre-established initial convolutional neural network based on sample rubber sleeve images, and the rubber sleeve fault detection result includes fault category information, confidence corresponding to the fault category information, and fault marking box position information.

[0028] Specifically, the target rubber sleeve image is input into a pre-trained rubber sleeve fault detection model. The rubber sleeve fault detection model identifies the fault category information in the target rubber sleeve image, determines the confidence level corresponding to the fault category information, and annotates the position area of ​​the identified faulty rubber sleeve through a labeling box to generate a model output result.

[0029] Optionally, the target rubber sleeve fault detection model includes an input layer, a backbone network, a feature enhancement layer and an output layer; the target rubber sleeve image is input into a pre-trained rubber sleeve fault detection model, and the rubber sleeve fault detection result is determined based on the model output result, including: transmitting the input target rubber sleeve image to the backbone network through the input layer, performing feature extraction on the target rubber sleeve image through the backbone network to obtain an original feature map of the rubber sleeve, and outputting the original feature map of the rubber sleeve to the feature enhancement layer; performing feature enhancement processing on the original feature map of the rubber sleeve through the feature enhancement layer, and transmitting the obtained rubber sleeve enhanced feature map to the output layer; performing fault prediction on the rubber sleeve enhanced feature map through the output layer to obtain fault category information and a confidence level corresponding to the fault category information, annotating the faulty rubber sleeve based on the prior frame to obtain fault annotation frame position information, and outputting the fault category information, the confidence level corresponding to the fault category information and the fault annotation frame position information as the model output result.

[0030] The prior frame can be understood as a preset rectangular annotation frame, which is used to locate the area where the faulty rubber sleeve is located.

[0031] Specifically, the target rubber sleeve image is input to the input layer of the pre-trained rubber sleeve fault detection model. The input target rubber sleeve image is subjected to feature extraction through the backbone network to obtain the original feature map of the rubber sleeve. The original feature map of the rubber sleeve is then transmitted to the feature enhancement layer, and the original feature map of the rubber sleeve is subjected to feature pooling processing to obtain enhanced feature data, and the enhanced feature data is subjected to feature fusion processing to obtain the enhanced feature map of the rubber sleeve. The enhanced feature map of the rubber sleeve is transmitted to the output layer for fault prediction. The output layer predicts whether there is a fault in the rubber sleeve and the specific category of the fault based on the enhanced features in the enhanced feature map of the rubber sleeve. At the same time, the output layer also provides a confidence level for each predicted fault category, indicating the confidence level of the model in the prediction. The detected faulty rubber sleeve is annotated based on the prior frame, and the location information of the fault annotation box is generated, that is, the specific location of the fault in the image. Finally, the model will output: fault category information (that is, the detected fault type), the confidence level corresponding to the fault category information (indicating the accuracy of the prediction), and the location information of the fault annotation box (indicating the location of the fault in the image).

[0032] Optionally, the backbone network includes multiple convolutional layers, non-linear activation function layers, attention modules and pooling layers, wherein the structure of the multiple convolutional layers includes an inverted residual block structure; the feature extraction of the target rubber sleeve image through the backbone network to obtain a rubber sleeve original feature map includes: performing point-by-point convolution and depth-separable convolution on the target rubber sleeve image through the convolutional layer to obtain a first rubber sleeve feature map, and performing point-by-point convolution on the first rubber sleeve feature map to obtain a second rubber sleeve feature map; performing nonlinear conversion on the second rubber sleeve feature map through the activation function layer, and adjusting the feature weight of the converted second rubber sleeve feature map through the attention module to obtain a third rubber sleeve feature map; reducing the spatial resolution of the third rubber sleeve feature map through the pooling layer to obtain the rubber sleeve original feature map.

[0033] Specifically, the target rubber sleeve image is convolved point by point through the convolution layer. The point by point convolution mainly changes the number of channels, and the convolution operation is performed independently on each channel through the depth separable convolution to reduce the amount of calculation to obtain the first rubber sleeve feature map. Then the first rubber sleeve feature map is convolved point by point to obtain the second rubber sleeve feature map. The target rubber sleeve image is first mapped to a high-dimensional space, expanded, and then restored to a low latitude. Exemplarily, assuming that the input feature is x, first perform a 1*1 convolution to expand the number of channels, then perform a depth separable convolution, and then restore the feature to a low dimension through a 1×1 convolution. Through a nonlinear activation function layer, such as a Hard-Swish activation function, the second rubber sleeve feature map is nonlinearly transformed to enhance the expression ability of the model. The attention module adjusts the weight of the third rubber sleeve feature map according to the importance of the feature, so that the model pays more attention to important features. Through a pooling layer, such as maximum pooling or average pooling, the spatial resolution of the third rubber sleeve feature map can be reduced, thereby extracting more representative features to form the original feature map of the rubber sleeve.

[0034] Optionally, before inputting the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, it also includes: obtaining a preset number of sample rubber sleeve images, for each sample rubber sleeve image, annotating the faulty rubber sleeve in each image based on the minimum circumscribed rectangle, and determining the annotation color of the minimum circumscribed rectangle based on the fault category corresponding to the faulty rubber sleeve to obtain an expected output image with a sample annotation box; constructing a sample data set based on the sample rubber sleeve image and the expected output image, and dividing the sample data set into a training set and a validation set; iteratively training a pre-constructed initial convolutional neural network model based on the training set, adjusting the model parameters through a back-propagation algorithm, and evaluating the model performance indicators of the model through a validation set after each training cycle; and determining the rubber sleeve fault detection model based on multiple model performance indicators.

[0035] Specifically, a preset number of sample rubber sleeve images are obtained from the actual production or test environment. Ensure that the sample rubber sleeve images cover all possible fault types and normal states. Use image annotation tools to manually or automatically annotate the faulty rubber sleeve area in each image, and generate an expected output image with a sample annotation box containing the coordinates of the minimum circumscribed rectangle. Assign a specific color or label to each fault category. When annotating the sample rubber sleeve image, ensure that the correct color or label is assigned to the minimum circumscribed rectangle according to the fault category. And determine the annotation file based on the annotated expected output image, and integrate the sample rubber sleeve image and its corresponding annotation file into a data set. Among them, the annotation file includes the label data of the fault category annotated for the sample rubber sleeve image, the center point coordinate data of the annotation box, the width data of the annotation box, and the height data of the annotation box. The data set is randomly divided into a training set and a validation set to ensure the generalization ability of the model training. Select a suitable initial convolutional neural network model according to the complexity of the task. Preprocess the training set (such as normalization, data enhancement, etc.), iteratively train the model using the training set, and adjust the model parameters through the back propagation algorithm. After each training cycle, the validation set is used to evaluate the performance indicators of the model (such as accuracy, recall, average precision, etc.). If the performance indicators on the validation set no longer improve, the training is stopped to avoid overfitting. The performance of the model is comprehensively evaluated based on multiple performance indicators. The model with the best performance is selected as the final rubber sleeve fault detection model.

[0036] Optionally, after constructing the sample data set based on the sample rubber cover images with sample annotation frames, the method further includes: respectively determining the size data and aspect ratio data of the sample annotation frames in each sample rubber cover image; clustering the size data and aspect ratio data of multiple sample annotation frames based on a clustering algorithm to obtain at least one prior frame; generating a prior frame list based on at least one prior frame, and integrating the prior frame list into a pre-constructed initial convolutional neural network model.

[0037] First, each image and its corresponding annotation data need to be read. The annotation data usually contains the bounding box coordinates of each target object (for example, the coordinates of the upper left corner and the lower right corner, or the center point coordinates and width and height). For each annotation box, calculate its width and height, and then calculate the aspect ratio (width / height). The width, height or aspect ratio of the annotation box is input into the clustering algorithm as a feature. The clustering algorithm divides multiple annotation boxes into K clusters, each cluster representing a group of annotation boxes with similar size and / or aspect ratio. The center of each cluster (i.e., the cluster center) can be regarded as a priori box. Each priori box corresponds to a common target size and shape. A priori box list is generated based on at least one priori box, and the priori box list is integrated into a pre-built initial convolutional neural network model.

[0038] Optionally, after obtaining a preset number of sample rubber cover images, the method further includes: preprocessing the sample rubber cover images to obtain preprocessed sample rubber cover images, wherein the preprocessing includes at least one of horizontal flipping, vertical flipping, image blurring, and hue and saturation transformation.

[0039] S130. Generate rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result, and display the rubber sleeve fault prompt information.

[0040] Exemplarily, based on preset fault features (such as cracks, breakage, discoloration, deformation and falling off, etc.), it is identified whether these fault features exist in the rubber sleeve fault detection result. If so, rubber sleeve fault prompt information corresponding to the target substation is generated based on the rubber sleeve fault detection result. The rubber sleeve fault prompt information is displayed through a preset notification method, such as a pop-up window, sound prompt or SMS notification, to ensure that the user can notice the fault prompt information in time.

[0041] The technical solution of the embodiment of the present invention is to obtain at least one target rubber sleeve image of the rubber sleeve of the target substation equipment; obtain the image of the rubber sleeve of the target substation equipment so as to perform fault detection based on the target rubber sleeve image later; then, input the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, and determine the rubber sleeve fault detection result based on the model output result; wherein, the rubber sleeve fault detection model is obtained by training a pre-established initial convolutional neural network based on the sample rubber sleeve image, and the rubber sleeve fault detection result includes fault category information, confidence corresponding to the fault category information, and fault marking box position information; perform fault detection on the target rubber sleeve image through the rubber sleeve fault detection model to improve detection efficiency and accuracy; finally, generate rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result, and display the rubber sleeve fault prompt information. The problem of low efficiency and low accuracy of rubber sleeve fault detection is solved, and the beneficial effect of improving the fault detection efficiency and accuracy of the rubber sleeve of the substation equipment is achieved.

[0042] Embodiment 2

[0043] Figure 2A flowchart of a fault detection method for a rubber sleeve of a substation equipment provided in the second embodiment of the present invention, this embodiment is a further optimization of how to generate rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result in the above embodiment. Optionally, the generation of rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result includes: determining the rubber sleeve fault level corresponding to the target substation equipment based on the fault category information and the confidence level corresponding to the fault category information, determining the rubber sleeve fault position corresponding to the target substation equipment based on the fault marking box position information; generating the rubber sleeve fault prompt information corresponding to the target substation equipment based on the rubber sleeve fault level and the rubber sleeve fault position.

[0044] like Figure 2 As shown, the method includes:

[0045] S210, obtaining at least one target rubber sleeve image of a target substation equipment rubber sleeve.

[0046] S220: Input the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, and determine a rubber sleeve fault detection result based on a model output result.

[0047] S230, determining the rubber sleeve fault level corresponding to the target substation equipment based on the fault category information and the confidence level corresponding to the fault category information, and determining the rubber sleeve fault position corresponding to the target substation equipment based on the fault marking box position information. The rubber sleeve fault detection result includes the fault category information, the confidence level corresponding to the fault category information, and the fault marking box position information;

[0048] The rubber sleeve fault level can be understood as the fault severity level of the rubber sleeve, and the rubber sleeve fault position can be understood as the position of the faulty rubber sleeve in the target rubber sleeve image.

[0049] Optionally, the rubber sleeve fault category information includes at least one of complete detachment of the rubber sleeve, partial detachment of the rubber sleeve, aging of the rubber sleeve, and damage of the rubber sleeve.

[0050] Specifically, clear levels are pre-set for fault categories based on business needs and safety standards. For example, "complete detachment of the rubber sleeve" is considered the highest level of serious fault, because it may directly lead to equipment failure or safety hazards; and "aging of the rubber sleeve" may be regarded as a lower level fault in the early stage, but as the degree of aging deepens, its level may also increase accordingly. Combine the fault category and confidence level to make a comprehensive judgment on the fault level. If the confidence level of a certain category is high and the corresponding fault level itself is very serious, it should be determined as a high-level fault. The location of the fault is accurately marked on the original image based on the fault annotation box position information. This helps operation and maintenance personnel to quickly locate the fault point and conduct subsequent inspections and repairs.

[0051] Preferably, in order to improve maintenance efficiency, visual assistance can be provided in the user interface, such as drawing a labeling box directly on the image and marking the fault level and location information, so that the operation and maintenance personnel can understand the fault situation at a glance.

[0052] S240: Generate rubber sleeve fault prompt information corresponding to the target substation equipment based on the rubber sleeve fault level and the rubber sleeve fault position, and display the rubber sleeve fault prompt information.

[0053] Specifically, the fault level, location information, and possible recommended measures (such as immediate replacement, regular inspection, etc.) are integrated into a complete fault report. The fault report is displayed to the user or related system through the user interface or remote communication means. The fault level and location information are combined to form a complete fault report. The fault report may include a description of the rubber sleeve fault category, an assessment of the rubber sleeve fault level, specific markings of the rubber sleeve fault location, and possible recommended measures. The fault report can be output and displayed in a variety of ways, such as printing it out for technicians to view, displaying it in real time on electronic equipment to patrol personnel, or sending it to a remote monitoring center through remote communication means.

[0054] The technical solution of the embodiment of the present invention determines the rubber sleeve fault level corresponding to the target substation equipment based on the fault category information and the confidence level corresponding to the fault category information. The fault type of the rubber sleeve can be accurately judged through the fault category information and the confidence level. The confidence level provides an assessment of the accuracy of the judgment, which helps to reduce false alarms and missed alarms and improve the reliability of diagnosis. The rubber sleeve fault position corresponding to the target substation equipment is determined based on the fault marking box position information; the fault marking box position information can accurately indicate the specific location of the rubber sleeve fault, which is convenient for maintenance personnel to quickly find and repair the fault. The maintenance time is reduced, the work efficiency is improved, and the power outage time caused by the fault is reduced. The rubber sleeve fault prompt information corresponding to the target substation equipment is generated based on the rubber sleeve fault level and the rubber sleeve fault position. The generated fault prompt information contains the fault level and location, and the maintenance personnel can quickly formulate a maintenance plan based on this information.

[0055] As an optional example of an embodiment of the present invention, the fault detection method of the rubber sleeve of the substation equipment in this embodiment specifically includes the following steps:

[0056] Step 1. Obtain a preset number of sample rubber sleeve images.

[0057] Step 2. Label the sample rubber sleeve image.

[0058] For each sample rubber sleeve image, annotate the faulty rubber sleeve in each image based on the minimum circumscribed rectangle, and determine the annotation color of the minimum circumscribed rectangle based on the fault category corresponding to the faulty rubber sleeve to obtain a desired output image with a sample annotation frame;

[0059] High-definition images that meet the use requirements are selected from the collected sample rubber sleeve images. Due to the complex environment of the substation and the changeable usage scenarios, the number of images of the rubber sleeve falling off state is limited. Therefore, in order to improve the generalization ability of the model, it is necessary to expand the image data. To achieve this goal, we applied a series of image enhancement techniques, including horizontal flipping, vertical flipping, image blurring, and random transformation of hue and saturation. These techniques can effectively increase the diversity of the data set, thereby reducing the dependence on large-scale labeled data. The horizontal and vertical flipping operations improve the model's ability to recognize targets under various viewing angles by simulating the device status in different directions; introducing a certain degree of image blurring helps the model adapt to the low-definition images that may appear in the substation, and enhances its ability to detect targets in blurred images; randomly adjusting the hue and saturation of the image to cope with changes under different lighting conditions enables the model to maintain good detection performance under various lighting environments. Through the above operations, the data set is expanded, the diversity and number of training samples are improved, and the robustness and accuracy of the model in different environments and conditions are significantly improved to meet the requirements of practical applications.

[0060] The substation equipment rubber sleeve images are manually annotated using the annotation system. The rubber sleeves in the image are annotated using the minimum circumscribed rectangle. The annotation box usually includes the entire rubber sleeve, but may also include other parts of a small part of the equipment. The mutual occlusion of equipment brings certain difficulties to the annotation. At the same time, different colors are used to annotate the rubber sleeves in different states. For example, the green box represents the normal rubber sleeve, the red box represents the rubber sleeve that has not completely fallen off, and the black box represents the rubber sleeve that has completely fallen off.

[0061] A sample data set is constructed based on the sample rubber sleeve image and the expected output image, and the sample data set is divided into a training set and a validation set.

[0062] Step 3. Integrate the prior box list into the pre-built initial convolutional neural network model.

[0063] The size data and aspect ratio data of the sample annotation frame in each sample rubber sleeve image are determined respectively; the size data and aspect ratio data of multiple sample annotation frames are clustered based on a clustering algorithm to obtain at least one prior frame; a prior frame list is generated based on the at least one prior frame, and the prior frame list is integrated into a pre-built initial convolutional neural network model.

[0064] Step 4. Model training.

[0065] The pre-built initial convolutional neural network model is iteratively trained based on the training set, the model parameters are adjusted through the back propagation algorithm, and the model performance indicators of the model are evaluated through the validation set after each training cycle; the rubber sleeve fault detection model is determined based on multiple model performance indicators.

[0066] Step 5. Deploy the trained model to the edge device to detect rubber sleeve failures. When the system detects a rubber sleeve failure, it issues a real-time warning.

[0067] Export the trained rubber sleeve fault detection model to a format suitable for edge devices, build the required operating environment on the edge device, install the necessary libraries and dependencies, and ensure that the software environment is compatible with the model. Develop a real-time monitoring system that integrates image acquisition, model reasoning, and result display. Design a suitable user interface so that maintenance personnel can intuitively view the detection results and prompt information. After deployment, regularly monitor the performance of the system, including detection accuracy, response time, and resource usage, to evaluate the performance of the model in actual applications.

[0068] The technical solution of the embodiment of the present invention can automatically identify and analyze the image of the target rubber sleeve, and quickly and accurately detect potential faults. Compared with the traditional manual detection method, this automated detection method greatly improves the accuracy and efficiency of fault detection, and reduces the interference and misjudgment of human factors. It realizes real-time fault monitoring of the rubber sleeve, and once an abnormality is found, a fault prompt is immediately issued. It helps to find and handle faults in time, prevent the faults from further deteriorating, and thus ensure the normal operation of the equipment and production safety.

[0069] Embodiment 3

[0070] Figure 3 This is a schematic diagram of the structure of a fault detection device for a rubber sleeve of a substation equipment provided by the third embodiment of the present invention. Figure 3 As shown, the device includes: an image acquisition module 310, a detection result determination module 320 and a prompt module 330.

[0071] Among them, the image acquisition module 310 is used to obtain at least one target rubber sleeve image of the rubber sleeve of the target substation equipment; the detection result determination module 320 is used to input the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, and determine the rubber sleeve fault detection result based on the model output result; wherein the rubber sleeve fault detection model is obtained by training a pre-established initial convolutional neural network based on the sample rubber sleeve image, and the rubber sleeve fault detection result includes fault category information, confidence corresponding to the fault category information, and fault marking box position information; the prompt module 330 is used to generate rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result, and display the rubber sleeve fault prompt information.

[0072] The technical solution of the embodiment of the present invention is to obtain at least one target rubber sleeve image of the rubber sleeve of the target substation equipment through the image acquisition module; obtain the image of the rubber sleeve of the target substation equipment so as to perform fault detection based on the target rubber sleeve image later; then, through the detection result determination module, input the target rubber sleeve image into the pre-trained rubber sleeve fault detection model, and determine the rubber sleeve fault detection result based on the model output result; wherein, the rubber sleeve fault detection model is obtained by training the pre-established initial convolutional neural network based on the sample rubber sleeve image, and the rubber sleeve fault detection result includes fault category information, confidence corresponding to the fault category information and fault marking box position information; perform fault detection on the target rubber sleeve image through the rubber sleeve fault detection module to improve the detection efficiency and accuracy; finally, through the prompt module, generate rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result, and display the rubber sleeve fault prompt information. The problem of low efficiency and low accuracy of rubber sleeve fault detection is solved, and the beneficial effect of improving the fault detection efficiency and accuracy of the rubber sleeve of the substation equipment is achieved.

[0073] Optionally, the target rubber sleeve fault detection model includes an input layer, a backbone network, a feature enhancement layer and an output layer; accordingly, the detection result determination module includes:

[0074] A feature extraction unit, used for transmitting the input target rubber sleeve image to the backbone network through the input layer, performing feature extraction on the target rubber sleeve image through the backbone network to obtain a rubber sleeve original feature map, and outputting the rubber sleeve original feature map to the feature enhancement layer;

[0075] A feature enhancement unit, used for performing feature enhancement processing on the original feature map of the rubber sleeve through the feature enhancement layer, and transmitting the obtained rubber sleeve enhanced feature map to the output layer;

[0076] A fault labeling unit is used to perform fault prediction on the rubber sleeve enhanced feature map through an output layer to obtain fault category information and a confidence level corresponding to the fault category information, label the faulty rubber sleeve based on a priori frame to obtain fault labeling frame position information, and output the fault category information, the confidence level corresponding to the fault category information, and the fault labeling frame position information as a model output result.

[0077] Optionally, the backbone network includes multiple convolutional layers, nonlinear activation function layers, attention modules and pooling layers, wherein the structure of the multiple convolutional layers includes an inverted residual block structure; accordingly, the feature extraction unit includes:

[0078] A convolution subunit, configured to perform point-by-point convolution and depth-wise separable convolution on the target rubber sleeve image through the convolution layer to obtain a first rubber sleeve feature map, and perform point-by-point convolution on the first rubber sleeve feature map to obtain a second rubber sleeve feature map;

[0079] A nonlinear conversion unit, used for performing nonlinear conversion on the second rubber sleeve feature map through an activation function layer, and adjusting a feature weight of the converted second rubber sleeve feature map through an attention module to obtain a third rubber sleeve feature map;

[0080] The original feature map acquisition subunit is used to reduce the spatial resolution of the third rubber sleeve feature map through the pooling layer to obtain the rubber sleeve original feature map.

[0081] Optionally, the device also includes a sample labeling module, a data set construction module, a training module and a model determination module.

[0082] The sample annotation module is used to obtain a preset number of sample rubber sleeve images before inputting the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, annotate the faulty rubber sleeve in each sample rubber sleeve image based on the minimum circumscribed rectangle, and determine the annotation color of the minimum circumscribed rectangle based on the fault category corresponding to the faulty rubber sleeve, so as to obtain a desired output image with a sample annotation frame;

[0083] The data set construction module is used to construct a sample data set based on the sample rubber sleeve image and the expected output image, and divide the sample data set into a training set and a verification set;

[0084] The training module is used to iteratively train the pre-built initial convolutional neural network model based on the training set, adjust the model parameters through the back propagation algorithm, and evaluate the model performance indicators of the model through the validation set after each training cycle;

[0085] The model determination module is used to determine the rubber sleeve fault detection model based on a plurality of the model performance indicators.

[0086] Optionally, the device further includes a labeling box associated data determination module, a priori box acquisition module, and a priori box integration module.

[0087] The annotation frame associated data determination module is used to respectively determine the size data and aspect ratio data of the sample annotation frame in each sample rubber sleeve image after the sample data set is constructed based on the sample rubber sleeve image with the sample annotation frame;

[0088] The prior frame acquisition module is used to cluster the size data and aspect ratio data of multiple sample annotation frames based on a clustering algorithm to obtain at least one prior frame;

[0089] The prior box integration module is used to generate a prior box list based on at least one prior box, and integrate the prior box list into a pre-built initial convolutional neural network model.

[0090] Optionally, the device also includes a preprocessing module.

[0091] Wherein, the preprocessing module is used to preprocess the sample rubber cover images after obtaining a preset number of sample rubber cover images to obtain preprocessed sample rubber cover images, wherein the preprocessing includes at least one of horizontal flipping, vertical flipping, image blurring, and hue and saturation transformation.

[0092] Optionally, the prompt module includes:

[0093] A fault information determination unit, configured to determine a rubber sleeve fault level corresponding to the target substation equipment based on the fault category information and the confidence level corresponding to the fault category information, and determine a rubber sleeve fault position corresponding to the target substation equipment based on the fault marking frame position information;

[0094] A fault prompt unit is used to generate rubber sleeve fault prompt information corresponding to the target substation equipment based on the rubber sleeve fault level and the rubber sleeve fault position.

[0095] Optionally, the rubber sleeve fault category information includes at least one of complete detachment of the rubber sleeve, partial detachment of the rubber sleeve, aging of the rubber sleeve, and damage of the rubber sleeve.

[0096] The fault detection device for a rubber sleeve of a power substation equipment provided in an embodiment of the present invention can execute the fault detection method for a rubber sleeve of a power substation equipment provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0097] Embodiment 4

[0098] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0099] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0100] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0101] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as the method for fault detection of rubber sleeves of substation equipment.

[0102] In some embodiments, the method for fault detection of the rubber sleeve of the substation equipment can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the method for fault detection of the rubber sleeve of the substation equipment described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the method for fault detection of the rubber sleeve of the substation equipment by any other appropriate means (for example, by means of firmware).

[0103] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0104] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0105] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0106] To provide interaction with a service acquirer, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the service acquirer; and a keyboard and a pointing device (e.g., a mouse or a trackball), through which the service acquirer can provide input to the electronic device. Other types of devices may also be used to provide interaction with a service acquirer; for example, the feedback provided to the service acquirer may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the service acquirer may be received in any form (including acoustic input, voice input, or tactile input).

[0107] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a service acquirer computer having a graphical service acquirer interface or a web browser through which the service acquirer can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0108] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0109] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0110] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A fault detection method for a rubber sleeve of a power substation, characterized in that: include: Acquire at least one target rubber sleeve image of a target substation equipment rubber sleeve; The target rubber sleeve image is input into a pre-trained rubber sleeve fault detection model, and a rubber sleeve fault detection result is determined based on the model output result; wherein the rubber sleeve fault detection model is obtained by training a pre-established initial convolutional neural network based on the sample rubber sleeve image, and the rubber sleeve fault detection result includes fault category information, confidence corresponding to the fault category information, and fault annotation box position information; Rubber sleeve fault prompt information corresponding to the target substation equipment is generated according to the rubber sleeve fault detection result, and the rubber sleeve fault prompt information is displayed.

2. The method according to claim 1, characterized in that The target rubber sleeve fault detection model includes an input layer, a backbone network, a feature enhancement layer and an output layer; the target rubber sleeve image is input into a pre-trained rubber sleeve fault detection model, and the rubber sleeve fault detection result is determined based on the model output result, including: The target rubber sleeve image is transmitted to the backbone network through the input layer, and features of the target rubber sleeve image are extracted through the backbone network to obtain a rubber sleeve original feature map, and the rubber sleeve original feature map is output to the feature enhancement layer; Performing feature enhancement processing on the original feature map of the rubber sleeve through the feature enhancement layer, and transmitting the obtained rubber sleeve enhanced feature map to the output layer; The fault prediction is performed on the rubber sleeve enhanced feature map through the output layer to obtain fault category information and the confidence corresponding to the fault category information, the faulty rubber sleeve is labeled based on the prior frame to obtain the fault labeling frame position information, and the fault category information, the confidence corresponding to the fault category information and the fault labeling frame position information are output as the model output result.

3. The method according to claim 2, characterized in that The backbone network includes multiple convolutional layers, nonlinear activation function layers, attention modules and pooling layers, wherein the structure of the multiple convolutional layers includes an inverted residual block structure; the feature extraction of the target rubber sleeve image through the backbone network to obtain the original feature map of the rubber sleeve includes: Performing point-by-point convolution and depth-wise separable convolution on the target rubber sleeve image through the convolution layer to obtain a first rubber sleeve feature map, and performing point-by-point convolution on the first rubber sleeve feature map to obtain a second rubber sleeve feature map; Performing a nonlinear transformation on the second rubber sleeve feature map through an activation function layer, and adjusting a feature weight of the transformed second rubber sleeve feature map through an attention module to obtain a third rubber sleeve feature map; The spatial resolution of the third rubber sleeve feature map is reduced by the pooling layer to obtain the rubber sleeve original feature map.

4. The method according to claim 1, characterized in that Before inputting the target rubber sleeve image into the pre-trained rubber sleeve fault detection model, the method further includes: Acquire a preset number of sample rubber sleeve images, and for each sample rubber sleeve image, annotate the faulty rubber sleeve in each image based on the minimum circumscribed rectangle, and determine the annotation color of the minimum circumscribed rectangle based on the fault category corresponding to the faulty rubber sleeve, so as to obtain a desired output image with a sample annotation frame; Constructing a sample data set based on the sample rubber sleeve image and the expected output image, and dividing the sample data set into a training set and a validation set; Iteratively training the pre-built initial convolutional neural network model based on the training set, adjusting the model parameters through a back-propagation algorithm, and evaluating the model performance indicators of the model through a validation set after each training cycle; The rubber sleeve fault detection model is determined based on a plurality of the model performance indicators.

5. The method according to claim 4, characterized in that After constructing the sample data set based on the sample rubber sleeve image with the sample annotation frame, the method further includes: Determine the size data and aspect ratio data of the sample annotation frame in each sample rubber sleeve image respectively; Clustering the size data and aspect ratio data of multiple sample annotation boxes based on a clustering algorithm to obtain at least one prior box; A priori box list is generated based on at least one prior box, and the prior box list is integrated into a pre-built initial convolutional neural network model.

6. The method according to claim 4, characterized in that After obtaining a preset number of sample rubber sleeve images, the method further includes: The sample rubber sleeve image is preprocessed to obtain a preprocessed sample rubber sleeve image, wherein the preprocessing includes at least one of horizontal flipping, vertical flipping, image blurring, and hue and saturation transformation.

7. The method according to claim 1, characterized in that The generating of rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result includes: Determine the rubber sleeve fault level corresponding to the target substation equipment based on the fault category information and the confidence level corresponding to the fault category information, and determine the rubber sleeve fault position corresponding to the target substation equipment based on the fault marking box position information; The rubber sleeve fault prompt information corresponding to the target substation equipment is generated based on the rubber sleeve fault level and the rubber sleeve fault position.

8. The method according to claim 1, characterized in that The rubber sleeve fault category information includes at least one of complete shedding of the rubber sleeve, partial shedding of the rubber sleeve, aging of the rubber sleeve, and damage of the rubber sleeve.

9. A fault detection device for a rubber sleeve of a power substation, characterized in that: include: An image acquisition module, used to acquire at least one target rubber sleeve image of a target substation equipment rubber sleeve; A detection result determination module is used to input the target rubber sleeve image into a pre-trained rubber sleeve fault detection model, and determine the rubber sleeve fault detection result based on the model output result; wherein the rubber sleeve fault detection model is obtained by training a pre-established initial convolutional neural network based on the sample rubber sleeve image, and the rubber sleeve fault detection result includes fault category information, confidence corresponding to the fault category information, and fault annotation box position information; A prompt module is used to generate rubber sleeve fault prompt information corresponding to the target substation equipment according to the rubber sleeve fault detection result, and display the rubber sleeve fault prompt information.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the fault detection method for a rubber sleeve of a substation equipment according to any one of claims 1 to 8 when executed.

Images