A substation periphery hidden danger change detection method, system, device and medium

Abstract

The application discloses a substation periphery hidden danger change detection method, system, equipment and medium, the method comprises the following steps: acquiring image data, identifying the nest building behavior mode in the image data; in response to the identification of the nest building behavior mode, feature extraction is carried out from the image data, and an enhanced nest body image is obtained; material type identification operation is carried out on the enhanced nest body image, and a material type classification result is obtained; the conductivity risk is judged according to the material type classification result, and a conductivity risk mark is obtained; early accurate discovery of hidden dangers is realized through nest building behavior mode identification, automatic judgment of the conductivity risk is realized through material type identification and conductivity attribute database query, quantitative evaluation of arc risk is realized through the fusion of conductivity risk, air gap, voltage level and environmental humidity and rainfall multiple factors, and the monitoring frequency is dynamically adjusted in combination with equipment influence evaluation, forming a complete closed loop from early identification, quantitative evaluation to differentiated prevention and control.

Description

Technical Field

[0001] This invention relates to the field of substation hazard detection technology, and in particular to a method, system, equipment and medium for detecting changes in hazards around a substation. Background Technology

[0002] Substations are the core hubs of power grid transmission and distribution, and their operational safety directly affects the stability of social electricity supply. The open spaces and suitable temperatures within substations, including portal frames and equipment supports, make them frequent nesting sites for birds. Materials commonly used by birds for nesting, such as metal wires and damp twigs, are conductive and can easily trigger arcing in the high-voltage electric field environment of substations (110kV and above). Furthermore, strong winds and thunderstorms can cause nests to detach, damaging equipment insulation components and causing significant losses, including power outages at the entire substation.

[0003] To address the aforementioned risks, existing technologies primarily employ manual inspections, bird deterrent devices, or surveillance cameras, but all have significant limitations. In early hazard identification, manual inspections are inefficient and time-consuming, making it difficult to detect early nesting behaviors such as bird site selection and material handling in real time. Surveillance equipment can only capture images and lacks intelligent analysis capabilities of bird activity characteristics, failing to distinguish nesting behavior from ordinary roosting or foraging. Even when nests are found, whether the materials contain conductive components such as metal fragments or damp branches still requires close manual inspection. However, inspections of high-voltage equipment areas pose a risk of electric shock and cannot dynamically assess material properties during nesting, resulting in long-term unmonitored development of hazards. Regarding risk assessment and response, existing technologies rely on a single dimension for risk judgment, depending solely on subjective human assessment, without considering key factors such as the voltage level of the power equipment, environmental parameters, and the distance between the nest and the equipment. This makes it impossible to accurately assess the probability of accidents such as arc discharge. The prevention and control measures lack specificity, and usually require power outages and cleaning after the nest has formed. This not only affects the reliability of power supply, but also makes it impossible to dynamically adjust the monitoring intensity according to the risk level. High-risk nests cannot be monitored more intensively, and low-risk nests cannot be monitored more leniently. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method, system, equipment, and medium for detecting changes in potential hazards around substations. This addresses the problem that existing technologies struggle to identify nesting behavior and the electrical conductivity risks of nesting materials in the early stages, and lack the ability to assess and differentiate prevention and control based on multiple factors such as voltage level and environmental parameters, thus remaining at the level of post-event emergency repairs.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for detecting changes in potential hazards around a substation, comprising: Acquire image data and identify nest-building behavior patterns within the image data; in response to the identification of the nest-building behavior patterns, extract features from the image data to obtain an enhanced nest image; perform material type identification on the enhanced nest image to obtain a material type classification result; determine the conductivity risk based on the material type classification result to obtain a conductivity risk marker; acquire relevant information about power equipment, assess the risk of arc discharge accidents based on the conductivity risk marker and the relevant information about power equipment, and generate a risk alarm signal; acquire an equipment impact assessment report based on the risk alarm signal; adjust the monitoring frequency based on the equipment impact assessment report to obtain the final monitoring configuration.

[0007] As a preferred embodiment of the substation perimeter hazard change detection method of the present invention, the step of identifying nest-building behavior patterns in the image data includes: preprocessing the image data to obtain a preprocessed image; extracting bird activity features from the preprocessed image; acquiring environmental parameter data, determining the bird activity level based on the environmental parameter data, and generating an active period marker; combining the bird activity features with the active period marker to form an activity feature set; filtering records that match the nest-building features through the activity feature set to determine nest-building behavior, thereby obtaining the nest-building behavior pattern.

[0008] As a preferred embodiment of the substation perimeter hazard change detection method of the present invention, the step of extracting bird activity features from the preprocessed image includes: calling a preset target detection algorithm to detect the preprocessed image to obtain the bird target bounding box coordinates; extracting the bird outline area, center point position and continuous inter-frame displacement distance based on the bird target bounding box coordinates; and determining the bird's dwell time at the center point position based on the continuous inter-frame displacement distance.

[0009] As a preferred embodiment of the substation perimeter hazard change detection method of the present invention, the step of feature extraction from the image data includes: locating the center point of the nest based on the position coordinates in the nest-building behavior pattern, constructing a detection box to extract a preliminary nest area; performing a morphological closing operation on the preliminary nest area to fill the internal cavities and fracture edges of the nest, obtaining a processed nest area; performing edge detection on the processed nest area using a preset edge detection algorithm to extract the nest outline point set, and calculating the coordinates of the circumscribed rectangle vertices of the outline point set; cropping the complete nest area from the image data based on the circumscribed rectangle vertices, recording the absolute coordinates of the cropped area in the original image as nest position information, obtaining a nest area image; and performing image enhancement operation on the nest area image to obtain the enhanced nest image.

[0010] As a preferred embodiment of the substation perimeter hazard change detection method of the present invention, the step of performing image enhancement operation on the nest area image includes: performing histogram equalization on the nest area image to redistribute pixel grayscale values ​​to enhance contrast, obtaining an equalized image; applying a bilateral filter to the equalized image to maintain edge sharpness and suppress noise, obtaining a filtered image; applying a Laplacian operator to the filtered image for convolution to extract detail components; and superimposing the detail components and the filtered image at the pixel level according to a preset weight coefficient to highlight the texture features of the nest material and the metallic reflective areas, obtaining the enhanced nest image.

[0011] As a preferred embodiment of the substation perimeter hazard change detection method of the present invention, the step of performing material type identification on the enhanced nest image includes: dividing the enhanced nest image into blocks according to a preset grid size to obtain several image blocks; extracting texture features from the image blocks and converting the image blocks from a first color space to a second color space, and extracting statistical feature values ​​of color components; combining the texture features and the statistical feature values ​​of color components to form a multidimensional feature vector; calculating the similarity of the multidimensional feature vectors between adjacent image blocks, and merging adjacent image blocks to obtain a material region in response to the similarity exceeding a preset similarity threshold; fusing the multidimensional feature vectors of the merged image blocks to obtain the multidimensional feature vector of the material region, and recording the position information of the material region in the enhanced nest image; analyzing the brightness component distribution characteristics of the material region, and generating a high reflectivity marker in response to the brightness value changing in space exceeding a preset amplitude threshold; inputting the multidimensional feature vector of the material region and the high reflectivity marker into a classifier to determine the material category and obtain the material type classification result.

[0012] As a preferred embodiment of the substation perimeter hazard change detection method of the present invention, the step of determining conductivity risk based on the material type classification result includes: querying a material conductivity property database based on the material type classification result to obtain the corresponding resistivity value; outputting a high conductivity risk indicator in response to the resistivity value being less than a preset threshold; outputting a low conductivity risk indicator in response to the resistivity value being greater than the preset threshold; maintaining the risk level unchanged in response to the conductivity risk indicator being a high conductivity risk indicator; correcting the risk level to medium conductivity risk in response to the conductivity risk indicator being a low conductivity risk indicator and the current ambient humidity value exceeding a preset humidity threshold; and combining and storing the risk level with the location information of the material area in the enhanced nest image to obtain a conductivity risk marker.

[0013] Secondly, the present invention provides a substation perimeter hazard change detection system, comprising: The system includes: an image acquisition module for acquiring image data; a nest-building behavior recognition module for recognizing nest-building behavior patterns in the image data; a feature extraction module for extracting features from the image data in response to the recognition of the nest-building behavior patterns to obtain an enhanced nest image; a material type recognition module for performing material type recognition on the enhanced nest image to obtain a material type classification result; a conductivity risk assessment module for judging conductivity risk based on the material type classification result to obtain a conductivity risk marker; an accident risk assessment module for acquiring relevant information about power equipment, assessing the risk of arc discharge accidents based on the conductivity risk marker and the relevant information about power equipment, and generating a risk alarm signal; an impact assessment module for obtaining an equipment impact assessment report based on the risk alarm signal; and a monitoring configuration module for adjusting the monitoring frequency based on the equipment impact assessment report to obtain the final monitoring configuration.

[0014] Thirdly, the present invention provides an electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method for detecting changes in hidden dangers around the substation are implemented.

[0015] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for detecting changes in potential hazards around a substation.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: early and accurate detection of potential hazards is achieved through nest-building behavior pattern recognition; automatic judgment of conductive risks is achieved through material type identification and conductivity attribute database query; quantitative assessment of arc risk is achieved by integrating multiple factors such as conductivity risk, air gap, voltage level, and environmental humidity and rainfall; and monitoring frequency is dynamically adjusted in conjunction with equipment impact assessment, forming a complete closed loop from early identification and quantitative assessment to differentiated prevention and control. This effectively realizes the pre-emptive prevention of conductive hazards caused by bird nests in substations and improves the safety and reliability of power system operation. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1This is a schematic diagram of the overall process of a method for detecting changes in potential hazards around a substation, according to an embodiment of the present invention.

[0019] Figure 2 This is a flowchart illustrating the steps for identifying nest-building behavior patterns in image data according to an embodiment of the present invention.

[0020] Figure 3 This is a flowchart illustrating the feature extraction steps for image data according to an embodiment of the present invention. Detailed Implementation

[0021] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0022] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for detecting changes in potential hazards around a substation is provided, comprising: Acquire image data and identify nest-building behavior patterns in the image data; In response to the recognition of the nest-building behavior pattern, feature extraction is performed from the image data to obtain an enhanced nest image; Perform material type identification on the enhanced nest image to obtain material type classification results; Based on the material type classification results, the conductivity risk is determined, and a conductivity risk label is obtained; Acquire relevant information about power equipment, assess the risk of arc discharge accidents based on the conductivity risk marker and the relevant information about power equipment, and generate a risk warning signal; Based on the aforementioned risk warning signals, obtain the equipment impact assessment report; The monitoring frequency is adjusted based on the equipment impact assessment report to obtain the final monitoring configuration.

[0023] As attached Figure 1As shown, existing technologies have two main shortcomings in preventing and controlling bird nest hazards in substations: In the early identification stage, manual inspections are inefficient and have long intervals, and monitoring equipment lacks intelligent analysis capabilities, making it impossible to distinguish between nesting behavior and ordinary roosting behavior. The risk of material conductivity requires close manual inspection, posing a risk of electric shock that cannot be dynamically assessed. In the risk assessment and prevention stage, risk judgment relies solely on subjective human experience, without considering key factors such as equipment voltage levels, environmental parameters, and nest distance. Furthermore, it lacks differentiated prevention capabilities and cannot dynamically adjust monitoring intensity according to risk levels, resulting in technical means remaining at the level of post-incident repair rather than proactive prevention.

[0024] Addressing the pain points of existing technologies for preventing bird nest hazards in substations, such as lack of early identification, subjective risk assessment, and passive and lagging prevention and control, this paper solves the problem of early detection from the source by intelligently identifying nest-building behavior patterns, solves the safety risks and efficiency issues of manual judgment by image recognition and database matching of material conductivity properties, achieves a leap from manual experience-based judgment to quantitative assessment through a multi-factor coupled risk assessment model, and realizes a transformation from passive response to proactive prevention through a risk-driven dynamic monitoring mechanism. In this way, the paper comprehensively improves the intelligence level and early warning timeliness of bird nest hazard prevention and control in substations.

[0025] Example 2, refer to Figure 1 As an embodiment of the present invention, based on the above embodiment, a method for detecting changes in potential hazards around a substation is provided, comprising: S1. Acquire image data and identify nest-building behavior patterns in the image data; Reference Figure 2 As shown, the process of identifying nest-building behavior patterns in the image data is presented. The steps for identifying nest-building behavior patterns in the image data include S1.1 to S1.5: S1.1. Preprocess the image data to obtain a preprocessed image; The image data is converted to grayscale, transforming the color image into a grayscale image; noise is removed from the grayscale image using a filtering method; the noise-removed image is then output as the preprocessed image. Specifically, the image data is converted to grayscale. Then, the acquired color image containing birds and equipment structure is converted to grayscale image to obtain grayscale image. Gaussian filtering is used to remove noise from grayscale image. Then, Gaussian kernel is used to perform weighted averaging of pixel values ​​in the neighborhood of each pixel in the image to suppress random noise introduced during image acquisition, resulting in preprocessed image.

[0026] S1.2 Extract bird activity features from the preprocessed image; The steps for extracting bird activity features from the preprocessed image include S1.2A~S12C: S1.2A. Call a preset target detection algorithm to detect the preprocessed image and obtain the coordinates of the bird target bounding box; Call the preset target detection algorithm and load the pre-trained model weights; input the pre-processed image into the target detection algorithm for detection; output the bounding box coordinates of the detected bird targets; Preferably, a preset object detection algorithm is invoked, and the weights of a pre-trained object detection model are loaded. This model has been fine-tuned using a substation bird dataset, which includes labeled images of magpies, crows, sparrows, and other birds. The pre-processed image is input into the object detection algorithm, which divides the image into multiple grids. Each grid predicts multiple bounding boxes and their confidence scores. After object detection, the algorithm outputs the coordinates of the detected bird bounding boxes, including the coordinates of the top-left corner (x, y). 1， y1) and the coordinates of the lower right corner (x) 2， y2).

[0027] S12B: Extract the bird outline area, center point position, and inter-frame displacement distance based on the bird target bounding box coordinates; The area of ​​the bounding box is calculated based on the coordinates of the bird target bounding box to obtain the bird outline area; the center point position of the bounding box is calculated based on the coordinates of the bird target bounding box to obtain the center point position; the spatial distance between the center point positions of adjacent frames is calculated to obtain the inter-frame displacement distance. Preferably, based on the coordinates of the top left corner (x) of the bird target bounding box. 1， y1) and the coordinates of the lower right corner (x) 2， The bird's outline area is calculated by (x1-x2)×(y1-y2); the center point position is calculated by ((x1+x2) / 2, (y1+y2) / 2); the spatial distance between the center points of adjacent frames is calculated by the Euclidean distance formula to obtain the displacement distance between consecutive frames.

[0028] S12C. Determine the duration of the bird's stay at the center point based on the continuous inter-frame displacement distance; The system determines whether the bird is in a stationary state at the center point based on the displacement distance between consecutive frames; the system accumulates the number of consecutive frames in a stationary state, and converts the accumulated stationary frame count into a time length based on the image acquisition frame rate to obtain the duration of the bird's stationary state at the center point. Specifically, the judgment is based on the displacement distance between consecutive frames. If the displacement distance between consecutive frames is less than the preset displacement threshold, it is determined that the bird is in a stationary state at the center point. The number of consecutive frames in the stationary state is accumulated, and the accumulated stationary frame number is converted into time length according to the image acquisition frame rate to obtain the stationary time of the bird at the center point.

[0029] S1.3. Obtain environmental parameter data, determine the activity level of birds based on the environmental parameter data, and generate active period markers; The system acquires environmental parameter data and determines the activity level of birds based on the temperature and humidity values ​​in the environmental parameter data. If the temperature value is within the preset temperature range and the humidity value is below the preset humidity threshold, it is marked as a high-activity period; otherwise, it is marked as a low-activity period, and an activity period marker is generated. The ambient temperature and humidity values ​​are collected in real time by temperature and humidity sensors deployed near the camera. The activity level of birds is judged based on the ambient temperature and humidity values. If the ambient temperature is within the preset temperature range and the humidity is below the preset humidity threshold, it can be marked as a high activity period. Otherwise, it is marked as a low activity period, and an activity period label is generated.

[0030] S1.4 Combine the bird activity characteristics with the active period markers to form an activity characteristic set; Specifically, the bird's outline area, center point position, inter-frame displacement distance, and dwell time are obtained as bird activity features, and active period markers are obtained. The bird's outline area, center point position, inter-frame displacement distance, and dwell time are associated and combined with the active period markers to finally form an activity feature set.

[0031] S1.5. Filter records that match the nest-building characteristics through the activity feature set, determine the nest-building behavior, and obtain the nest-building behavior pattern; Preferably, bird activity records with a dwell time exceeding a preset time threshold are selected based on the activity feature set. The number of round-trip flights between two fixed locations with the bird carrying the object is counted. If the number of round-trip flights exceeds a preset number threshold, the starting coordinates and ending coordinates of the round-trip path are extracted. The cumulative increment of the object is calculated based on the change in the outline area of ​​the object carried in consecutive frames. If the cumulative increment of the object is continuously positive and the ending coordinates remain unchanged, it is determined to be nest building behavior. The nest building location coordinates, nest building start time, material handling frequency and voltage level data, and equipment geographical coordinate information are associated and stored to obtain the nest building behavior pattern.

[0032] S2. In response to the recognition of the nest-building behavior pattern, feature extraction is performed from the image data to obtain an enhanced nest image; Reference Figure 3 As shown, the process of obtaining the enhanced nest image from the image data by feature extraction is presented. The steps of feature extraction from the image data include S2.1 to S2.5: S2.1 Locate the center point of the nest based on the position coordinates in the nest building behavior pattern, and construct a detection box to extract the preliminary nest area; The location coordinates are read from the nest-building behavior pattern, and the center point of the nest is marked in the image data. A detection box is generated based on the center point. The detection box is used to extract a local image region containing the nest from the image data as a preliminary nest region for subsequent morphological processing. Preferably, the center point of the nest is located in the image data based on the position coordinates in the nest building behavior pattern, and a rectangular detection box with a side length of a preset ratio of the image width is constructed based on the center point of the nest. The image area within the detection box is then extracted as the preliminary nest area.

[0033] S2.2 Perform morphological closing operation on the preliminary nest area to fill the internal cavities and broken edges of the nest, and obtain the processed nest area; A morphological closing operation is performed on the preliminary nest region. This closing operation consists of a combination of dilation and erosion operations, with dilation followed by erosion. The dilation operation can fill the voids inside the nest caused by sparse branches, while the erosion operation can repair the breaks at the edge of the nest caused by incomplete image segmentation, so that the nest outline is connected and complete, resulting in the processed nest region. Specifically, a morphological closing operation is performed on the preliminary nest area. Using a circular structural element of a preset size, an expansion operation is first performed on the preliminary nest area to fill the internal cavity of the nest, and then an erosion operation is performed to repair the broken edges, so that the nest outline is connected and complete, resulting in the processed nest area.

[0034] S2.3. Perform edge detection on the processed nest area using a preset edge detection algorithm, extract the nest outline point set, and calculate the coordinates of the vertices of the bounding rectangle of the outline point set. The pre-defined edge detection algorithm is invoked to perform edge detection on the processed nest region. The edge detection algorithm identifies the position of pixels with drastic gray-scale changes by calculating the gradient magnitude and gradient direction of each pixel in the image, and connects the pixels that meet the conditions to form a contour. Based on the detection results, the contour point set constituting the nest boundary is output. The contour point set is fitted with the minimum bounding rectangle to obtain the coordinates of the vertices of the bounding rectangle. Specifically, the pre-defined Canny edge detection algorithm is invoked to perform edge detection on the processed nest region. The Sobel operator is used to calculate the gradient components in the horizontal and vertical directions and synthesize the gradient magnitude and direction. Local gradient maximum points are retained through non-maximum suppression. The nest contour point set is output by using double threshold detection and edge connection processing. The nest contour point set is fitted with a minimum bounding rectangle to obtain the minimum bounding rectangle of the nest contour point set. The coordinates of the four vertices of the minimum bounding rectangle are calculated and used as the vertex coordinates of the bounding rectangle.

[0035] S2.4. Based on the coordinates of the circumscribed rectangle vertices, crop the complete nest area from the image data, record the absolute coordinates of the cropped area in the original image as the nest position information, and obtain the nest area image. The complete nest region is located in the image data using the coordinates of the circumscribed rectangle vertices as the clipping boundary; the rectangular region is clipped from the image data to obtain the image of the complete nest region; at the same time, the absolute pixel coordinates of the top left vertex of the rectangular region in the original image data are recorded as nest position information for subsequent positioning steps. Preferably, the cropping range containing the complete nest area is located in the image data based on the coordinates of the circumscribed rectangle vertices. A rectangular area containing the complete nest area is cropped from the image data. The absolute coordinates of the top-left vertex of the cropping range in the image data are recorded as the nest location information. The image of the cropped complete nest area is then output as the nest area image.

[0036] S2.5. Perform image enhancement operation on the nest area image to obtain the enhanced nest image.

[0037] The steps for feature extraction from the image data include S2.5A to S2.5D: S2.5A. Perform histogram equalization on the image of the nest region to redistribute pixel gray values ​​to enhance contrast and obtain an equalized image. The frequency of each gray level pixel in the nest region image is counted to form a gray level histogram; the cumulative distribution function is calculated based on the histogram, and the mapping relationship from the original gray level value to the new gray level value is established using the cumulative distribution function; the original gray level value of each pixel is converted into a new gray level value according to the mapping relationship, so that the gray level distribution of the image changes from concentrated to uniform, enhancing the image contrast and obtaining an equalized image. Histogram equalization is performed on the nest region image. The gray-level histogram of the nest region image is statistically analyzed, the cumulative distribution function is calculated, and the original gray-level values ​​are mapped to the entire gray-level range according to the cumulative distribution function, so that the pixel gray-level values ​​are redistributed, the image contrast is enhanced, and an equalized image is obtained.

[0038] S2.5B. A bilateral filter is used to filter the equalized image to maintain edge sharpness and suppress noise, resulting in a filtered image. A bilateral filter is used to filter the equalized image. In the spatial domain, the filter uses a Gaussian kernel to assign smaller weights to pixels that are far from the center pixel, and in the pixel value domain, it uses a Gaussian kernel to assign smaller weights to pixels with large gray-level differences. While smoothing noise in uniform areas of the image, it retains information about gray-level abrupt changes at the edges and outputs a filtered image. Specifically, a bilateral filter is used to filter the equalized image. The spatial domain Gaussian kernel of the bilateral filter assigns lower weights to pixels that are far apart to suppress the influence of spatial distance. The pixel value domain Gaussian kernel of the bilateral filter assigns lower weights to pixels with large gray-level differences to protect gray-level jumps at the edges. While smoothing noise, the clarity of the nest material boundary is maintained, resulting in a filtered image.

[0039] S2.5C. Apply the Laplacian operator to the filtered image to perform convolution operation and extract detail components; A Laplacian operator convolution kernel is constructed, which detects the location of abrupt changes in grayscale values ​​in the image through second derivative operations. The Laplacian operator convolution kernel is then used to perform sliding convolution operations on the filtered image to extract high-frequency detail components that reflect the texture details, edge contours, and metallic reflective boundaries of the nest material. Specifically, the Laplacian operator is applied to the filtered image for convolution. The second derivative operation of the Laplacian operator is used to highlight the locations in the filtered image where the gray values ​​change drastically, and the high-frequency detail components in the filtered image are extracted as detail components.

[0040] S2.5D: The detailed components and the filtered image are superimposed pixel-level according to a preset weight coefficient to highlight the texture features of the nest material and the metallic reflective areas, thereby obtaining the enhanced nest image.

[0041] Specifically, the process involves acquiring detailed components and a filtered image; then, according to a preset weighting coefficient, multiplying the gray value of the detailed component by the weighting coefficient at each pixel location and adding it to the gray value of the filtered image to achieve pixel-level overlay; through overlay, fine textures, material boundaries, and metallic reflective areas that were originally inconspicuous in the filtered image are enhanced, resulting in an enhanced nest image.

[0042] S3. Perform material type identification on the enhanced nest image to obtain the material type classification result; The steps for performing material type identification on the enhanced nest image include S3.1 to S3.6: S3.1. Divide the enhanced nest image into blocks according to a preset grid size to obtain several image blocks; Acquire an enhanced nest image; divide the enhanced nest image into uniform horizontal and vertical sections according to a preset grid size; treat each sub-region as an independent image block and output several image blocks for subsequent feature extraction. Preferably, the enhanced nest image is divided into blocks according to a preset grid size, and the enhanced nest image is evenly divided into several sub-regions in the horizontal and vertical directions. Each sub-region is an independent image block, resulting in several image blocks.

[0043] S3.2 Extract texture features from the image patch, convert the image patch from the first color space to the second color space, and extract statistical feature values ​​of color components; For each image patch, a texture descriptor is extracted; the gray value distribution relationship of adjacent pixel pairs within the image patch in different directions is statistically analyzed to obtain texture features that reflect texture uniformity, complexity, and contrast; at the same time, the image patch is converted from the first color space to the second color space, and the statistical feature values ​​of each color component are extracted in the second color space as color features. Preferably, a gray-level co-occurrence matrix is ​​extracted for the image patch, and the frequency of occurrence of gray-level value pairs with a distance of 1 pixel in the four directions of 0 degrees, 45 degrees, 90 degrees, and 135 degrees is statistically analyzed. The second-order angular moment, entropy value, and moment of inertia are extracted from the gray-level co-occurrence matrix as texture descriptors. At the same time, the image patch is converted from the RGB color space to the HSV color space, and the statistical feature values ​​of the hue component, saturation component, and lightness component are extracted in the HSV color space, including the mean and standard deviation of each component, as statistical feature values ​​of the color components.

[0044] S3.3. Combine the texture features with the statistical feature values ​​of the color components to form a multidimensional feature vector; Preferably, the texture features of the image patch are obtained; the statistical feature values ​​of the color components of the image patch are obtained; the texture features and the statistical feature values ​​of the color components are concatenated along the feature dimension and merged into a multi-dimensional feature vector, which serves as the multi-dimensional feature vector of the image patch.

[0045] S3.4 Calculate the similarity of the multidimensional feature vectors between adjacent image blocks. In response to the similarity exceeding a preset similarity threshold, merge the adjacent image blocks to obtain a material region. Fuse the multidimensional feature vectors of the merged image blocks to obtain the multidimensional feature vector of the material region. Record the position information of the material region in the enhanced nest image. Obtain multidimensional feature vectors of adjacent image blocks; calculate the similarity between the multidimensional feature vectors of adjacent image blocks. The higher the similarity, the higher the probability that the two image blocks belong to the same type of material; compare the calculated similarity with a preset similarity threshold. If the similarity exceeds the preset similarity threshold, merge the adjacent image blocks into the same material region and output the material region. Preferably, the cosine similarity of the multidimensional feature vectors between adjacent image blocks is calculated, and the calculated cosine similarity is compared with a preset similarity threshold. If the cosine similarity exceeds the preset similarity threshold, a region growing algorithm is used to merge the adjacent image blocks into the same material region to obtain the material region.

[0046] S3.5 Analyze the lightness component distribution characteristics of the material area, and generate a high reflectivity mark in response to the change of lightness value in space exceeding a preset amplitude threshold. Obtain the lightness component of the material area in the second color space; analyze the variation range of the lightness value in the material area space, and calculate the standard deviation or range of the lightness value as a measure of the variation range; compare the variation range of the lightness value in space with the preset range threshold. If the variation range exceeds the preset range threshold, it indicates that the surface of the material area has specular reflection characteristics, and a high reflectivity mark is generated. Preferably, the distribution characteristics of the lightness component in the HSV color space of the material area are analyzed, the standard deviation of the lightness value in the material area space is calculated, and the calculated standard deviation is compared with a preset amplitude threshold. If the standard deviation exceeds the preset amplitude threshold, the material area is determined to have high reflectivity, and a high reflectivity mark is generated.

[0047] S3.6 Input the multidimensional feature vector of the material region and the high reflectivity marker into the classifier to determine the material category and obtain the material type classification result; Obtain the multidimensional feature vector of the material region; obtain the high reflectivity label of the material region; combine the multidimensional feature vector and the high reflectivity label and input them into the classifier; the classifier makes a judgment based on the distribution position of the feature vector in the feature space and outputs the material category label of the material region to obtain the material type classification result; Preferably, the multidimensional feature vector and high reflectivity label of the material region are input into the classifier. The classifier adopts a support vector machine with radial basis function kernel. The classifier makes a judgment based on the distribution position of the multidimensional feature vector and high reflectivity label in the feature space, and outputs the material category label to obtain the material type classification result.

[0048] S4. Determine the conductivity risk based on the material type classification results and obtain a conductivity risk label; The steps for determining conductivity risk based on the material type classification results include S4.1~S4.5: S4.1. Query the material conductivity property database according to the material type classification results to obtain the corresponding resistivity value; Obtain the material type classification results; use the material category in the material type classification results as the query index to access the preset material conductivity property database; match the resistivity value record corresponding to the material category in the database, and output the queried resistivity value; Specifically, the system uses the material category in the material type classification results as the query index to access the preset material conductivity property database. The material conductivity property database is divided into metal, organic and synthetic material categories. Each material category stores a range of resistivity values. The system matches the resistivity value record corresponding to the material category in the database to obtain the corresponding resistivity value.

[0049] S4.2. In response to the resistivity value being less than a preset threshold, output a high conductivity risk indicator; in response to the resistivity value being greater than the preset threshold, output a low conductivity risk indicator. Obtain the queried resistivity value; obtain a preset resistivity threshold; compare the resistivity value with the preset threshold; if the resistivity value is less than the preset threshold, determine that the material has a high conductivity risk and output a high conductivity risk flag; if the resistivity value is greater than the preset threshold, determine that the material has a low conductivity risk and output a low conductivity risk flag. Specifically, the resistivity value is compared with a preset threshold, which is set to 10^6 ohms·m according to the substation safety regulations. If the resistivity value is less than 10^6 ohms·m, the material is determined to have a high conductivity risk, and a high conductivity risk indicator is output. If the resistivity value is greater than 10^6 ohms·m, the material is determined to have a low conductivity risk, and a low conductivity risk indicator is output.

[0050] S4.3. In response to the conductivity risk identifier being a high conductivity risk identifier, the risk level remains unchanged; Specifically, the process involves obtaining the conductivity risk identifier; determining the category of the conductivity risk identifier; and if the conductivity risk identifier is a high conductivity risk identifier, then classifying the material's risk level as high risk and maintaining the risk level unchanged. S4.4 In response to the conductivity risk indicator being a low conductivity risk indicator and the current ambient humidity value exceeding a preset humidity threshold, the risk level is corrected to medium conductivity risk. Obtain the conductivity risk indicator; obtain the current ambient humidity value; determine whether the conductivity risk indicator is a low conductivity risk indicator and whether the current ambient humidity value exceeds the preset humidity threshold; if both conditions are met, dynamically adjust the risk judgment result according to the material's moisture absorption characteristic curve, and correct the material's risk level from medium risk level to medium conductivity risk; if the conditions are not met, no correction is made. Specifically, the system obtains the conductivity risk indicator and the current ambient humidity value, determines whether the conductivity risk indicator is a low conductivity risk indicator and whether the current ambient humidity value exceeds the preset humidity threshold, which is set to 85% according to the substation safety regulations; if the conductivity risk indicator is a low conductivity risk indicator and the current ambient humidity value exceeds 85%, then a dynamic risk assessment is performed based on the material's moisture absorption characteristic curve: taking dry wood material as an example, its resistivity in the dry state can reach 10^9 ohm-meters. When the ambient humidity exceeds 85%, the wood fibers absorb moisture to form conductive channels, and the resistivity drops to below 10^7 ohm-meters. At this time, the risk level of the material is corrected from medium risk level to medium conductivity risk.

[0051] S4.5 Combine and store the risk level with the location information of the material region in the enhanced nest image to obtain a conductive risk marker.

[0052] Obtain the final determined risk level; obtain the location information of the material region in the enhanced nest image; associate and combine the risk level with the location information of the material region in the enhanced nest image to form structured conductive risk marker data; output and store the conductive risk markers; Specifically, the final determined risk level is associated with the location information of the material region in the enhanced nest image to form structured conductive risk marker data containing material type, initial risk level, humidity-corrected risk level, and material spatial coordinates. The conductive risk markers are then stored in a database to obtain conductive risk markers.

[0053] S5. Obtain relevant information about the power equipment, assess the risk of arc discharge accidents based on the conductivity risk marker and the relevant information about the power equipment, and generate a risk alarm signal; The system acquires information related to power equipment, including voltage level data; calculates the air gap distance between the nest and the nearest power equipment based on the nest location coordinates in the conductivity risk marker and nest building behavior pattern; obtains the critical safety distance for the corresponding voltage level based on the voltage level data; if the air gap distance is less than the critical safety distance, calculates the ratio of the air gap distance to the critical safety distance to obtain a discharge probability index; assesses the percentage decrease in air insulation strength based on humidity and rainfall values ​​in the environmental parameter data, and corrects the discharge probability index to obtain a discharge risk value; if the discharge risk value is higher than a preset high-risk threshold, a high-risk alarm signal containing location information and risk level is generated; if the discharge risk value is lower than the preset high-risk threshold, a low-risk monitoring signal is generated. Specifically, the substation SCADA system reads real-time voltage level data of each power device to obtain relevant information, including voltage level data. Based on conductivity risk markers and nest location coordinates in nesting behavior patterns, it calculates the air gap distance between the nest and the nearest power device. Based on the voltage level data, it obtains the critical safety distance for the corresponding voltage level according to the minimum electrical clearance standards specified in electrical safety regulations. The critical safety distance for 110kV equipment is 1 meter, for 220kV equipment it is 2 meters, and for 500kV equipment it is 5 meters. If the air gap distance is less than 70% of the critical safety distance, the ratio of the air gap distance to the critical safety distance is calculated to obtain a discharge probability index. The system also assesses the percentage decrease in air insulation strength based on humidity and rainfall values ​​in environmental parameters. For dry air, relative humidity below 6% is considered a critical safety distance. The breakdown voltage remains stable at 0% humidity. When the humidity rises to 70%~80%, the insulation strength decreases by 15%~20%. When the humidity exceeds 85%, the insulation strength decreases by 30%~40%. Under rainfall conditions, raindrops elongate and deform under the influence of the electric field, further reducing the breakdown voltage. A weighted comprehensive evaluation method is used to correct the discharge probability index to obtain the discharge risk value. The weight coefficient of the discharge probability index is 0.5, the weight coefficient of the environmental humidity influence factor is 0.3, and the weight coefficient of the rainfall influence factor is 0.2. When the humidity value exceeds the benchmark value by 10%, the risk value increases by 0.1, and when the rainfall increases by 10 mm per hour, the risk value increases by 0.15. If the discharge risk value is higher than the preset high-risk threshold of 0.7, a high-risk alarm signal containing nest location information, equipment number information, and risk level information is generated. If the discharge risk value is lower than the preset high-risk threshold of 0.7, a low-risk monitoring signal is generated.

[0054] S6. Based on the risk alarm signal, obtain the equipment impact assessment report; Extract the location coordinates and risk level of the nest from the risk alarm signal; read the equipment number, coordinates, and connection relationship data from the power equipment location information; calculate the set of power equipment within a preset radius centered on the nest location, identify all downstream equipment on the fault propagation path by tracing the electrical connection path between equipment, and determine the preliminary list of affected equipment; for each equipment in the preliminary list of affected equipment, extract the current temperature and wind speed values ​​from the environmental parameter data; obtain the current allowable current carrying capacity by querying the equipment rated current carrying capacity correction coefficient table based on the temperature value; calculate the load rate by comparing the real-time load current of the equipment with the current allowable current carrying capacity, and mark the equipment as a high-risk operating state if the load rate exceeds the preset safety threshold, thus obtaining the equipment operating risk level; associate the equipment operating risk level with the equipment power supply range data, and count the user types and numbers within the power supply range of each equipment; calculate the degree of impact score by accumulating the different weight coefficients of residential users, industrial users, and important users; integrate the equipment name, risk level, number of affected equipment, statistics of affected users, and degree of impact score to form a structured data record, and obtain the equipment impact assessment report; Specifically, the system extracts the location coordinates and risk level of the fault nest based on the risk alarm signal, and reads the equipment number, coordinates, and connection relationship data from the power equipment location information. It calculates the set of power equipment within a preset radius centered on the fault nest location, traverses the equipment connection diagram using a breadth-first search algorithm to trace the electrical connection paths between equipment, marks affected equipment layer by layer starting from the fault point, identifies all downstream equipment on the fault propagation path, and determines a preliminary list of affected equipment. For each device in the preliminary list of affected equipment, it extracts the current temperature and wind speed values ​​from environmental parameter data. Based on the temperature value, it queries the equipment's rated current carrying capacity correction coefficient table. For every 10 degrees Celsius increase in ambient temperature, the allowable current carrying capacity of the equipment decreases by approximately 8%. Substation equipment is designed for an ambient temperature of 40 degrees Celsius; when the actual temperature is 30 degrees Celsius, the current carrying capacity can be increased by 8%, and when the temperature reaches 45 degrees Celsius, the current carrying capacity needs to be reduced by 4%. Increasing the wind speed from 0.5 m / s to 2 m / s can improve the convective heat dissipation coefficient. The equipment's current-carrying capacity is increased by 50% to obtain the current allowable current carrying capacity. The real-time load current of the equipment is compared with the current allowable current carrying capacity to calculate the load rate. If the load rate exceeds the preset safety threshold, it is marked as a high-risk operating state, and the equipment's operating risk level is obtained. The equipment's operating risk level is correlated with the equipment's power supply range data, and the user types and numbers within the power supply range of each equipment are counted. According to different weight coefficients for residential users, industrial users, and important users: the weight coefficient for first-level important users, including hospitals, government agencies, and communication hubs, is set to 10; the weight coefficient for industrial users is between 3 and 7 based on production continuity requirements; and the weight coefficient for ordinary residential users is 1. The impact score is calculated by summing these values. The equipment name, risk level, number of affected equipment, statistics of affected users, and impact score are integrated to form a structured data record containing four modules: basic equipment information, risk assessment results, statistics of impact range, and emergency response suggestions, resulting in an equipment impact assessment report.

[0055] S7. Adjust the monitoring frequency according to the equipment impact assessment report to obtain the final monitoring configuration; Extract the impact score and nest location information from the equipment impact assessment report; query the current monitoring frequency of sensors in the nest area; if the impact score exceeds the preset high impact threshold, increase the monitoring frequency to a preset multiple of the current value; if the impact score is lower than the preset high impact threshold, maintain the current monitoring frequency; write the adjusted monitoring frequency into the configuration parameters of the corresponding sensor and update the data acquisition cycle; record the frequency adjustment time and nest identifier to form a configuration record containing sensor number, monitoring frequency, and update time, thus obtaining the final monitoring configuration; Specifically, based on the equipment impact assessment report, the impact score and nest location information are extracted. The current monitoring frequency of the sensors in the nest's area is then queried; the current monitoring frequency is once every 5 minutes. If the impact score exceeds a preset high impact threshold of 0.7, the monitoring frequency is increased from once every 5 minutes to once every 1 minute, i.e., 5 times the current value. If the impact score is below the preset high impact threshold of 0.7, the current monitoring frequency of once every 5 minutes remains unchanged. The adjusted monitoring frequency is written to the configuration parameters of the corresponding sensor via the Modbus protocol, with the sensor's holding register address set to 0x0010. The data acquisition cycle is updated to match the new frequency. The frequency adjustment time and nest identifier are recorded, forming a configuration record containing the sensor number, monitoring frequency, and update time. This record is stored in a time-series database, resulting in the final monitoring configuration.

[0056] Example 3 illustrates a schematic scheme for a method of detecting changes in hazards around a substation. It should be noted that the technical solution of this system for detecting changes in hazards around a substation is based on the same concept as the technical solution of the method described above. Details not described in detail in this embodiment can be found in the description of the technical solution of the method described above.

[0057] This embodiment also provides a substation perimeter hazard change detection system, including: The system includes: an image acquisition module for acquiring image data; a nest-building behavior recognition module for recognizing nest-building behavior patterns in the image data; a feature extraction module for extracting features from the image data in response to the recognition of the nest-building behavior patterns to obtain an enhanced nest image; a material type recognition module for performing material type recognition on the enhanced nest image to obtain a material type classification result; a conductivity risk assessment module for judging conductivity risk based on the material type classification result to obtain a conductivity risk marker; an accident risk assessment module for acquiring relevant information about power equipment, assessing the risk of arc discharge accidents based on the conductivity risk marker and the relevant information about power equipment, and generating a risk alarm signal; an impact assessment module for obtaining an equipment impact assessment report based on the risk alarm signal; and a monitoring configuration module for adjusting the monitoring frequency based on the equipment impact assessment report to obtain the final monitoring configuration.

[0058] This embodiment also provides an electronic device suitable for detecting changes in potential hazards around substations, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the method for detecting changes in potential hazards around substations as proposed in the above embodiment.

[0059] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for detecting changes in potential hazards around a substation as proposed in the above embodiments.

[0060] The storage medium proposed in this embodiment and the method for detecting changes in hidden dangers around substations proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0061] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0062] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting changes in potential hazards around a substation, characterized in that, include: Acquire image data and identify nest-building behavior patterns in the image data; In response to the recognition of the nest-building behavior pattern, feature extraction is performed from the image data to obtain an enhanced nest image; Perform material type identification on the enhanced nest image to obtain material type classification results; Based on the material type classification results, the conductivity risk is determined, and a conductivity risk label is obtained; Acquire relevant information about power equipment, assess the risk of arc discharge accidents based on the conductivity risk marker and the relevant information about power equipment, and generate a risk warning signal; Based on the aforementioned risk warning signals, obtain the equipment impact assessment report; The monitoring frequency is adjusted based on the equipment impact assessment report to obtain the final monitoring configuration.

2. The method for detecting changes in potential hazards around a substation as described in claim 1, characterized in that, The steps for identifying nest-building behavior patterns in the image data include, The image data is preprocessed to obtain a preprocessed image; Extract bird activity features from the preprocessed image; Acquire environmental parameter data, determine bird activity levels based on the environmental parameter data, and generate active period markers; The bird activity characteristics are combined with the active period markers to form an activity characteristic set; By filtering records that match the nest-building characteristics using the activity feature set, nest-building behavior is determined, and the nest-building behavior pattern is obtained.

3. The method for detecting changes in potential hazards around a substation as described in claim 2, characterized in that, The steps for extracting bird activity features from the preprocessed image include: The preprocessed image is detected by calling a preset target detection algorithm to obtain the coordinates of the bird target bounding box; Extract the bird's outline area, center point position, and inter-frame displacement distance based on the bird target bounding box coordinates; The duration of the bird's stay at the center point is determined based on the continuous inter-frame displacement distance.

4. The method for detecting changes in potential hazards around a substation as described in claim 1, characterized in that, The steps for feature extraction from the image data include, Based on the location coordinates in the nest building behavior pattern, locate the center point of the nest and construct a detection box to extract the preliminary nest area; A morphological closing operation is performed on the preliminary nest area to fill the internal cavities and broken edges of the nest, resulting in the processed nest area. The processed nest region is edge-detected by a preset edge detection algorithm, the nest contour point set is extracted, and the coordinates of the vertices of the bounding rectangle of the contour point set are calculated. The complete nest region is cropped from the image data based on the coordinates of the circumscribed rectangle vertices, and the absolute coordinates of the cropped region in the original image are recorded as the nest position information to obtain the nest region image. An image enhancement operation is performed on the image of the nest region to obtain the enhanced nest image.

5. The method for detecting changes in potential hazards around a substation as described in claim 4, characterized in that, The steps for performing image enhancement on the nest region image include: Histogram equalization is performed on the image of the nest region to redistribute pixel gray values ​​to enhance contrast, resulting in an equalized image. A bilateral filter is used to filter the equalized image to maintain edge sharpness and suppress noise, resulting in a filtered image. The Laplacian operator is applied to the filtered image for convolution to extract detail components; The detailed components are superimposed on the filtered image at the pixel level according to a preset weighting coefficient to highlight the texture features of the nest material and the metallic reflective areas, thus obtaining the enhanced nest image.

6. The method for detecting changes in potential hazards around a substation as described in claim 1, characterized in that, The steps for performing material type identification on the enhanced nest image include: The enhanced nest image is divided into blocks according to a preset grid size to obtain several image blocks; Extract texture features from the image patch, and convert the image patch from the first color space to the second color space to extract statistical feature values ​​of color components; The texture features are combined with the statistical feature values ​​of the color components to form a multidimensional feature vector; Calculate the similarity of the multidimensional feature vectors between adjacent image blocks. In response to the similarity exceeding a preset similarity threshold, merge the adjacent image blocks to obtain a material region. Fuse the multidimensional feature vectors of the merged image blocks to obtain the multidimensional feature vector of the material region. Record the position information of the material region in the enhanced nest image. The lightness component distribution characteristics of the material area are analyzed, and a high reflectivity mark is generated in response to the spatial variation of the lightness value exceeding a preset amplitude threshold. The multidimensional feature vector of the material region and the high reflectivity label are input into the classifier to determine the material category and obtain the material type classification result.

7. The method for detecting changes in potential hazards around a substation as described in claim 6, characterized in that, The steps for determining conductivity risk based on the material type classification results include: Based on the material type classification results, query the material conductivity property database to obtain the corresponding resistivity value; In response to the resistivity value being less than a preset threshold, a high conductivity risk indicator is output; In response to the resistivity value being greater than the preset threshold, a low conductivity risk indicator is output. In response to the conductivity risk identifier being a high conductivity risk identifier, the risk level remains unchanged; In response to the fact that the conductivity risk indicator is a low conductivity risk indicator and the current ambient humidity value exceeds a preset humidity threshold, the risk level is corrected to medium conductivity risk; The risk level is combined with the location information of the material region in the enhanced nest image and stored to obtain a conductive risk marker.

8. A substation perimeter hazard change detection system, employing the method described in any one of claims 1-7, characterized in that, include: The image acquisition module is used to acquire image data; Nest building behavior recognition module, used to identify nest building behavior patterns in the image data; The feature extraction module is used to extract features from the image data in response to the recognition of the nest building behavior pattern, so as to obtain an enhanced nest image; The material type identification module is used to perform material type identification on the enhanced nest image to obtain the material type classification result. The conductivity risk assessment module is used to determine the conductivity risk based on the material type classification results and obtain a conductivity risk label. The accident risk assessment module is used to acquire relevant information about power equipment, assess the risk of arc discharge accidents based on the conductivity risk marker and the relevant information about power equipment, and generate a risk alarm signal. The impact assessment module is used to obtain an equipment impact assessment report based on the risk alarm signal; The monitoring configuration module is used to adjust the monitoring frequency based on the equipment impact assessment report to obtain the final monitoring configuration.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the substation periphery hazard change detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the substation perimeter hazard change detection method according to any one of claims 1 to 7.

Images