Substation defect automatic tracking and patrolling system based on visual analysis

Abstract

The invention discloses a transformer substation defect automatic tracking and patrolling system based on visual analysis, and the system comprises a front-end data collection module which is configured to collect image information of a monitoring area; the intelligent analysis module is configured to analyze the image information through computer vision so as to carry out defect detection and carry out abnormity identification based on a degree learning algorithm; the task management module is configured to generate alarm information and send the alarm information to an alarm module when defects or abnormities are detected; and the back-end management module is configured to store and manage the image data, the analysis result, the historical record and the inspection task data. According to the system, the advanced visual analysis technology is integrated, so that automatic identification of defects of power equipment and abnormal field environment is realized. Once a problem is detected, the system automatically generates a tracking patrol task, and timely notifies operation and maintenance personnel through an alarm pop-up window at a system end. In addition, the system can also automatically call related cameras to carry out high-frequency tracking inspection.

Description

Technical Field

[0001] This invention relates to a substation inspection system, and more particularly to a substation defect automatic tracking and inspection system based on visual analysis. Background Technology

[0002] Currently, intelligent substation equipment defect detection systems are widely used and researched in the industry. These systems primarily focus on efficiently identifying and analyzing defects in power equipment, automatically detecting equipment anomalies and potential problems through image recognition, deep learning, and artificial intelligence technologies. In recent years, with the improvement of computer hardware computing power and the development of related technologies, the development of computer vision technology has, to some extent, improved the equipment detection capabilities of intelligent substations. It is generally agreed that modern computer vision systems can achieve real-time monitoring and image analysis of substation sites, especially demonstrating high accuracy in image understanding and object recognition. Of course, in defect detection, there are also applications using deep learning algorithms to detect insulator damage, conductor slack, and equipment corrosion. Compared to traditional methods, deep learning models can automatically learn image features, thereby significantly improving detection accuracy. However, as intelligent substation systems develop towards integration, miniaturization, and collaboration, some problems have gradually emerged in currently common systems, such as: 1. Insufficient detection accuracy and robustness. Substation environments are typically complex and variable, with diverse equipment types, complex locations, and severe obstructions, which affects the system's recognition performance. In particular, adverse environments such as nighttime and rainy / snowy weather significantly reduce detection accuracy due to changes in lighting and weather conditions. Furthermore, existing systems rely on large amounts of labeled data for training, but due to the unique environment of substations, there is a lack of sufficient high-quality defect image datasets, limiting the model's generalization ability and making it prone to misjudgments or missed detections in unknown scenarios.

[0003] 2. Real-time performance and computing resource requirements High-precision deep learning models (such as ResNet and YOLO) typically involve large computational demands. When deployed on resource-constrained edge devices, this can lead to processing delays, failing to meet real-time detection requirements. Furthermore, substation monitoring cameras often generate massive amounts of video data, and real-time analysis and transmission of this data place high demands on system bandwidth and computing power. Under conditions of limited bandwidth or network instability, the system may fail to identify and issue alerts in a timely manner.

[0004] 3. Lack of multimodal fusion Most existing systems rely on a single visual sensor (such as an optical camera), which cannot fully utilize the advantages of other sensors such as infrared and lidar when facing complex scenes, resulting in limited defect detection accuracy.

[0005] 4. Model interpretability and maintenance issues Deep learning models, during the detection process, cannot explain why a certain judgment result occurs, making it difficult for maintenance personnel to understand the model's detection logic, thus affecting the system's reliability in practical applications. Furthermore, substation equipment changes over time, such as aging or being updated, and the detection models in the existing system need continuous updates and optimizations to adapt to new scenarios, a complex and time-consuming process.

[0006] 5. Low efficiency in alarm management and task scheduling. The system typically only issues simple alarms or notifications after detecting defects, lacking intelligent task scheduling and management. For example, in multi-area defect detection scenarios, the system cannot intelligently prioritize patrol tasks or allocate resources, potentially leading to delayed responses. This not only wastes human resources but may also affect the efficiency of diagnosing and handling actual faults. This shows that there is still room for improvement in the existing technology. Summary of the Invention

[0007] In view of the above-mentioned problems existing in the prior art, one aspect of the present invention is to provide a substation defect automatic tracking and inspection system based on visual analysis with higher accuracy and robustness.

[0008] To achieve the above objectives, one embodiment of the present invention provides an automatic substation defect tracking and inspection system based on visual analysis, comprising: The front-end data acquisition module is configured to acquire image information of the monitored area. The intelligent analysis module is configured to analyze the image information using computer vision to detect defects and to identify anomalies based on a degree learning algorithm. The task management module is configured to generate alarm information and send it to an alarm module when a defect or anomaly is detected. The backend management module is configured to store and manage image data, analysis results, historical records, and inspection task data.

[0009] Preferably, the front-end data acquisition module includes a high-definition camera, a drone, and an inspection robot.

[0010] Preferably, the intelligent analysis module uses the YOLO model when performing defect detection. During training, the model first collects and labels image data of common defects in substations (such as insulator damage, wire slack, equipment corrosion, etc.); then, it uses transfer learning to fine-tune the YOLO model pre-trained on the COCO large-scale dataset.

[0011] Preferably, the YOLO model used by the intelligent analysis module for defect detection is trained by combining multi-scale detection and feature pyramid network (FPN) to improve the model's ability to detect defects at different scales.

[0012] Preferably, the intelligent analysis module uses a convolutional neural network (CNN) combined with a time series model (such as LSTM or GRU) for anomaly detection.

[0013] Preferably, the system also includes a GPU-accelerated server, which has multiple computing nodes deployed, each of which has a convolutional neural network algorithm model deployed.

[0014] Preferably, the intelligent analysis module includes: The preprocessing module is configured to preprocess the acquired video and image data, including but not limited to image denoising, enhancement, and resolution adjustment. A feature extraction module is configured to extract edge features, morphological features, or color features of device defects from the preprocessed image information. The visual analysis module is configured to identify target devices in an image based on the edge features, shape features, or color features of device defects, and to locate the specific location of the target in the image.

[0015] Preferably, the intelligent analysis module further includes a feature fusion module, which is configured to extract the gradient direction histogram features and color histogram features of the equipment in the substation equipment defect detection, and form a fused feature vector as the input of the convolutional neural network algorithm model.

[0016] Preferably, the task management module includes: The task creation module is configured to generate high-frequency tracking and inspection tasks when an anomaly or defect is detected. The task execution module is configured to assign the front-end data acquisition module to perform inspection tasks.

[0017] Preferably, the task management module also includes a task tracking module, which is configured to track the progress of defects and update task status and inspection results.

[0018] Preferably, the backend management module includes: The data storage module is configured to store image data, detection results, and abnormal screenshots collected by the front-end data acquisition module. The historical data display module is configured to visually display the progress trends of historical detection records and equipment anomaly records. The system management module is configured to manage system users and logs. This invention relates to an automatic defect tracking and inspection system for substations. This system integrates advanced visual analysis technology to automatically identify defects in power equipment and anomalies in the on-site environment. The core of the system is an intelligent analysis module that processes image data from various areas of the substation in real time, identifying potential defects or anomalies through a preset algorithm model. Once a problem is detected, the system automatically generates a tracking and inspection task and promptly notifies maintenance personnel via an alarm pop-up window on the system end. Furthermore, the system can automatically call relevant cameras for high-frequency tracking and inspection until the defect or anomaly is confirmed to be resolved. The multimodal data fusion of data from various front-end data acquisition modules in this invention effectively improves detection accuracy and robustness. Attached Figure Description

[0019] Figure 1 This is a system structure block diagram of the substation defect automatic tracking and inspection system based on visual analysis according to the present invention.

[0020] Figure 2 This is a flowchart of the automatic substation defect tracking and inspection system based on visual analysis according to the present invention.

[0021] Key reference numerals: 100 - Front-end data acquisition module; 101 - High-definition camera; 102 - Drone; 103 - Inspection robot; 200 - Intelligent Analysis Module; 201 - Preprocessing Module; 202 - Feature Extraction Module; 203 - Visual Analysis Module; 300 - Task Management Module; 301 - Task Creation Module; 302 - Task Execution Module; 303 - Task Tracking Module; 400 - Backend Management Module; 401 - Data Storage Module; 402 - Historical Data Display Module; 403 - System Management Module; 500 Alarm Module; 600 - GPU-accelerated server, 601 - compute node. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0023] Obviously, the described embodiments are only a part of the embodiments of this disclosure, not all of them. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0024] like Figures 1 to 2 As shown, an embodiment of the present invention provides an automatic substation defect tracking and inspection system based on visual analysis, comprising: The front-end data acquisition module 100 is configured to acquire image information of the monitored area. The front-end data acquisition module 100 includes a high-definition camera, a drone, and an inspection robot. The high-definition camera needs to support high resolution (4K), infrared functionality, and a wide-angle view for all-weather monitoring. The inspection robot is equipped with a high-definition camera and an infrared thermal imager for ground inspection. The drone has infrared imaging, a high-definition camera, and stable hovering capabilities for high-altitude inspection.

[0025] The intelligent analysis module 200 is configured to analyze the image information using computer vision for defect detection and to perform anomaly recognition based on a degree learning algorithm. When performing defect detection, the intelligent analysis module 200 uses a YOLO model. During training, this model first collects and labels image data of common defects in substations (such as insulator damage, wire slack, equipment corrosion, etc.); then, it uses transfer learning to fine-tune the YOLO model pre-trained on the COCO large-scale dataset. During training, the YOLO model used for defect detection combines multi-scale detection and a feature pyramid network (FPN) to improve the model's ability to detect defects at different scales. Defect detection typically includes equipment surface defect detection, such as corrosion, cracks, oil stains, and deformation. Environmental anomaly detection includes foreign objects, water leakage, fire smoke, and abnormal temperatures. Further, in some embodiments, the intelligent analysis module uses a convolutional neural network (CNN) combined with a time series model (such as LSTM or GRU) for anomaly recognition.

[0026] In other improvements, such as Figure 1 As shown, the system also includes a GPU-accelerated server 600, which is equipped with multiple computing nodes 601 for model training and real-time analysis. Each computing node is equipped with a convolutional neural network algorithm model.

[0027] Specifically, in this invention, the intelligent analysis module 200 includes: The preprocessing module 201 is configured to preprocess the acquired video and image data. This preprocessing includes, but is not limited to, image denoising, enhancement, and resolution adjustment. Basic processing of the acquired image or video frames may also include correction and cropping, with the aim of improving image quality and providing better input for subsequent detection and recognition algorithms. When denoising images, methods such as Gaussian filtering and median filtering can be used to remove image noise. When enhancing images, histogram equalization and CLAHE (adaptive histogram equalization) can be used to improve image contrast. In addition, image correction typically includes geometric transformations and distortion correction based on conventional algorithms.

[0028] Feature extraction module 202 is configured to extract edge features, morphological features or color features of device defects from the preprocessed image information; The visual analysis module 204 is configured to identify target equipment in an image based on the edge features, morphological features, or color features of equipment defects, and to locate the specific position of the target in the image. As an improvement, the intelligent analysis module 200 also includes a feature fusion module 203, configured to extract the gradient direction histogram features and color histogram features of the equipment in substation equipment defect detection, forming a fused feature vector as input to the convolutional neural network algorithm model. Alternatively, a deep learning model can be used to extract convolutional features, which are then combined with traditional texture analysis features (such as LBP features), predicted separately by a classifier, and their predicted probabilities fused to obtain the final classification result. Alternatively, an attention mechanism can be used to selectively focus on detailed areas such as equipment cracks and corrosion to improve detection accuracy. Specifically, introducing an attention mechanism into the defect detection model can improve detection accuracy in the following ways: Local Focusing: Attention mechanisms can automatically determine which regions might contain defects based on the response intensity of feature maps. During forward propagation, the model can assign higher weights to small cracks or corrosion areas on the device surface, allowing the model to focus on these details and thus enhancing its defect detection capabilities. Alternatively, Feature enhancement: By adaptively adjusting the weights at different locations in the feature map, the attention mechanism can enhance the feature representation of potential defect regions. For example, for a tiny crack on the surface of a transformer, the attention mechanism can enhance the features of that region, making it easier for subsequent classification or detection modules to identify the anomaly.

[0029] For example, suppose there's an infrared image of substation equipment where an abnormal temperature is observed (possibly a hotspot caused by corrosion). In a typical CNN model, features from all regions are treated equally, potentially leading to a weak response to abnormal areas. However, by introducing an attention mechanism, the model automatically identifies the temperature anomaly area and focuses on it, thereby improving detection accuracy.

[0030] The system of this invention also includes a task management module 300, configured to generate alarm information and send it to an alarm module 500 when a defect or anomaly is detected; specifically, after a defect or anomaly is detected, alarm information is pushed based on real-time communication protocols such as WebSocket. The task management module 300 includes: a task creation module 301, configured to generate high-frequency tracking and inspection tasks when an anomaly or defect is detected; and a task execution module 302, configured to assign inspection tasks to a front-end data acquisition module. Preferably, the task management module 300 also includes a task tracking module 303, configured to track defect progress and update task status and inspection results.

[0031] The backend management module 400 is configured to store and manage image data, analysis results, historical records, and inspection task data. The backend management module 400 includes: a data storage module 401, configured to store image data, detection results, and anomaly screenshots collected by the frontend data acquisition module; a historical data display module 402, configured to visually display historical detection records and the progress trend of equipment anomaly records; and a system management module 403, configured to manage system users and logs.

[0032] Figure 2 This diagram illustrates the workflow of the visual analysis-based automatic substation defect tracking and inspection system of the present invention. Figure 2 As shown, the workflow of the system of the present invention includes: Data acquisition: Cameras and inspection equipment collect image data and transmit it to the intelligent analysis module.

[0033] Data Analysis: The intelligent analysis module processes images in real time to identify potential defects or anomalies.

[0034] Alarms and Task Generation: After a defect is detected, the task management module generates an inspection task and notifies the operation and maintenance personnel through the system.

[0035] Tracking and Inspection: The system automatically calls cameras or inspection equipment to conduct tracking and inspection, continuously monitoring problem areas.

[0036] Results storage and analysis: Inspection results and patrol data are stored in the backend database for historical analysis and trend prediction.

[0037] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0038] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0039] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0040] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0041] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0042] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0043] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0044] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0045] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0046] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A substation defect automatic tracking and inspection system based on visual analysis, including: The front-end data acquisition module is configured to acquire image information of the monitored area. The intelligent analysis module is configured to analyze the image information using computer vision to detect defects and to identify anomalies based on a degree learning algorithm. The task management module is configured to generate alarm information and send it to an alarm module when a defect or anomaly is detected. The backend management module is configured to store and manage image data, analysis results, historical records, and inspection task data.

2. The substation defect automatic tracking and inspection system based on visual analysis as described in claim 1, wherein the front-end data acquisition module includes a high-definition camera, a drone, and an inspection robot.

3. The automatic substation defect tracking and inspection system based on visual analysis as described in claim 1, wherein the intelligent analysis module uses the YOLO model when performing defect detection, and during the training of the model, image data of common defects in the substation are first collected and labeled; then transfer learning is used to fine-tune the YOLO model pre-trained on the COCO large-scale dataset.

4. In the substation defect automatic tracking and inspection system based on visual analysis as described in claim 3, the YOLO model used by the intelligent analysis module for defect detection is trained by combining multi-scale detection and feature pyramid network to improve the model's ability to detect defects at different scales.

5. The substation defect automatic tracking and inspection system based on visual analysis as described in claim 1, wherein the intelligent analysis module uses a convolutional neural network combined with a time series model for detection when performing anomaly identification.

6. The substation defect automatic tracking and inspection system based on visual analysis as described in claim 1 further includes a GPU-accelerated server, wherein the GPU-accelerated server is deployed with multiple computing nodes, and each computing node is deployed with a convolutional neural network algorithm model.

7. The automatic substation defect tracking and inspection system based on visual analysis as described in claim 1, wherein the intelligent analysis module comprises: The preprocessing module is configured to preprocess the acquired video and image data, including but not limited to image denoising, enhancement, and resolution adjustment. A feature extraction module is configured to extract edge features, morphological features, or color features of device defects from the preprocessed image information. The visual analysis module is configured to identify target devices in an image based on the edge features, shape features, or color features of device defects, and to locate the specific location of the target in the image.

8. The automatic substation defect tracking and inspection system based on visual analysis as described in claim 7, wherein the intelligent analysis module further includes a feature fusion module, wherein the feature fusion module is configured to extract the gradient direction histogram features and color histogram features of the equipment in the substation equipment defect detection, and form a fused feature vector as the input of the convolutional neural network algorithm model.

9. The substation defect automatic tracking and inspection system based on visual analysis as described in claim 1, wherein the task management module includes: The task creation module is configured to generate a tracking and inspection task when an anomaly or defect is detected. The task execution module is configured to assign the front-end data acquisition module to perform inspection tasks.

10. The substation defect automatic tracking and inspection system based on visual analysis as described in claim 1, wherein the task management module further includes a task tracking module configured to track defect progress and update task status and inspection results.

Images