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

Abstract

The application discloses a substation hidden danger change detection method, system, device and medium, and the method comprises. In response to the calculation of the sun vector of the second phase, a simulated shadow map is constructed, a first-level screening mechanism is performed on the simulated shadow map and the original change map, and a real change pixel set is screened out; the real change pixel set is compared with the collected spectral feature vector set of the first phase, and a real high-reflectivity change pixel set is obtained. The application eliminates the negative brightness pixels falling into the shadow area by calculating the sun vector and constructing the simulated shadow map, thereby guaranteeing the detection authenticity; and then, the spectral feature comparison and reflectivity determination are performed on the real change pixel set, and the hidden danger is subdivided into high-reflectivity change and ground darkening. The application effectively reduces the false alarm rate, clearly defines the physical properties of the hidden danger, realizes the rapid positioning of the change area to specific equipment, and improves the accuracy and practicability of the substation hidden danger monitoring.

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 substation hazards. Background Technology

[0002] In recent years, satellite remote sensing technology has been gradually applied to the field of power facility monitoring due to its advantages such as wide coverage, short revisit period, and relatively low cost. Existing technologies for substation hazard detection based on multi-source satellite data typically involve acquiring multi-temporal high-resolution optical satellite imagery and synthetic aperture radar (SAR) satellite data covering the substation area. Optical imagery is used for spectral information analysis and texture feature extraction to identify changes in surface cover. Simultaneously, differential interferometry (DII) is employed to process SAR data to quantitatively invert minute surface deformations. Finally, the analysis results from both datasets are combined to determine the location and type of the hazard.

[0003] However, when processing multi-temporal images, existing technologies typically perform pixel-by-pixel differential comparisons, failing to effectively distinguish between the device's own shadow movement and actual changes in land cover. Current technologies often rely on differences in spectral or textural features to analyze areas of change, determining whether an area has brightened or darkened, but without assigning a clear physical meaning. This prevents the detection results from directly linking to specific risks, lacks clear guidance for subsequent hazard investigation, and leads to inefficient response.

[0004] Therefore, there is a need for a substation hazard change detection method that can solve the technical problems in existing technologies, such as high false alarm rate due to shadow movement, unclear change attributes that cannot guide the handling, and the disconnect between detection results and equipment records, to meet practical needs. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a method, system, equipment and medium for detecting changes in substation hazards to solve the problems in the prior art, such as high false alarm rate due to shadow movement, unclear change attributes that cannot guide the handling, and the disconnect between detection results and equipment records.

[0007] 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 in substations, comprising: In response to the calculation of the solar vector in the second time phase, a simulated shadow map is constructed; Obtain the original change map of the second time phase compared with the first time phase, and perform a first-level filtering mechanism on the simulated shadow map and the original change map. The first-level filtering mechanism is used to filter out the set of real change pixels. The true set of changed pixels is compared with the set of spectral feature vectors collected in the first time phase to obtain the true set of highly reflective changed pixels. The original variation map is stitched together with the real set of high reflectivity variation pixels to generate a hazard attribute classification map; The hazard attribute classification map is clustered by adjacent spaces to generate change patches, and the change patch attribute table is obtained through the parameters of the change patches; By associating the attribute table of the changed patches with the equipment parameters of the substation, a list of potential hazard changes can be established.

[0008] As a preferred embodiment of the substation hazard change detection method described in this invention, wherein: The steps to construct a simulated shadow map include: Acquire the image information features of the second time phase and retrieve the geographic coordinates of the device under test; The solar position parameters relative to the device under test are obtained by performing function calculations on the image information features and the geographic coordinates of the device under test. The shadow distribution data is obtained by simulating the relationship between illumination and shadow projection based on the solar position parameters. The shadow distribution data is rendered to generate a simulated shadow map.

[0009] As a preferred embodiment of the substation hazard change detection method described in this invention, wherein: In response to the extraction of image information features from the first time phase and image information features from the second time phase, a feature mapping relationship is established; The image information features of the first time phase and the image information features of the second time phase are aligned according to the feature mapping relationship to obtain an intermediate map. The brightness difference value at each position is calculated using the intermediate spectrum, and the original variation spectrum is generated based on all the brightness difference values.

[0010] As a preferred embodiment of the substation hazard change detection method described in this invention, wherein:

[0011] Pre-identify shaded and non-shaded areas within the simulated shadow map; The positions with a brightness difference value less than 0 in the original change map are taken as negative pixels, and the positions with a brightness difference value greater than 0 are taken as positive pixels, thus generating a set of negative pixels and a set of positive pixels. In the set of negative pixels, query the position corresponding to each position in the simulated shadow map, and use the obtained position as the feature position; If a feature is located in a shaded area, it is removed from the negative pixel set; if a feature is located in a non-shaded area, it is retained in the negative pixel set. The negative pixel set and the positive pixel set are merged to form the true change pixel set.

[0012] As a preferred embodiment of the substation hazard change detection method described in this invention, wherein:

[0013] The steps for comparing the actual set of changed pixels with the set of spectral feature vectors acquired in the first time phase include: The image information features of the first time phase are divided into a preset background category set, and the first average spectral feature vector corresponding to each category in the background category set is calculated; Combine all the first average spectral eigenvectors into a first spectral eigenvector set; A two-level filtering mechanism is executed based on the real change pixel set, and the result of the two-level filtering mechanism is divided into a brightened pixel set and a darkened pixel set. Obtain the corresponding brightness spectral features of each position in the brightness pixel set in the image information features of the second time phase, and generate a brightness spectral feature set; The set of brightened spectral features is compared with the set of first spectral feature vectors to obtain the comparison value; A three-level screening mechanism is executed based on the comparison values, and a set of real high reflectivity variation pixels is generated at the positions corresponding to the results of the three-level screening mechanism.

[0014] As a preferred embodiment of the substation hazard change detection method described in this invention, wherein: The steps for stitching the original variation map with the actual set of highly reflective variation pixels include: Create a blank matrix and import all positions corresponding to the real high reflectivity change pixel set and the darkening pixel set into the blank matrix to generate a preliminary parameter matrix; Assign first category values ​​to all positions corresponding to the set of real high reflectivity variation pixels in the preliminary parameter matrix; Assign second category values ​​to all positions corresponding to the set of darkened pixels in the initial parameter matrix; Based on the preliminary parameter matrix that has been assigned values ​​for the first and second categories, a hazard attribute classification map is generated.

[0015] As a preferred embodiment of the substation hazard change detection method described in this invention, wherein:

[0016] The steps of generating change patches and obtaining a change patch attribute table based on the parameters of the change patches include: Identify adjacent locations with the same category value in the hazard attribute classification map, and merge the identified locations to generate change patches; Extract the coordinate data corresponding to the location of each changed patch, and calculate the mean of the coordinates. As the coordinates of the geometric center of each changing patch; Spatial resolution is obtained from the image information features of the first or second time phase. Based on the obtained spatial resolution and the number of locations in each changed patch, the area of ​​each changed patch is calculated. Extract the elevation data corresponding to the location in each changed patch, calculate the average elevation, and obtain the average elevation of each changed patch; In response to the sum of the geometric center coordinates, area, and average elevation of each changed feature, an attribute table of the changed feature is generated.

[0017] Secondly, the present invention provides a substation hazard change detection system, comprising:

[0018] The simulated shadow construction module is used to construct a simulated shadow map in response to the calculation of the solar vector in the second time phase; The original change map acquisition module is used to acquire the original change map of the second time phase compared with the first time phase; The first-level filtering module is used to perform a first-level filtering mechanism on the simulated shadow map and the original change map, and uses the result of the first-level filtering mechanism as the set of real change pixels. The spectral comparison module is used to compare the true set of changed pixels with the spectral feature vector set of the first time phase to obtain the true set of highly reflective changed pixels. The classification map generation module is used to stitch the original change map with the real set of high reflectivity change pixels to generate a hazard attribute classification map; The image clustering module is used to perform adjacent spatial clustering on the hidden danger attribute classification map, generate change images, and obtain the change image attribute table through the parameters of the change images. The list creation module is used to associate the changed patch attribute table with the equipment parameters of the substation to create a list of potential hazard changes.

[0019] 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 substation hidden danger change detection method are implemented.

[0020] 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 substation hazard change detection method.

[0021] Compared with existing technologies, the beneficial effects of this invention are as follows: By calculating the solar vector and constructing a simulated shadow map, this invention eliminates negative brightness pixels falling into the shadow area, ensuring the authenticity of the detection; furthermore, it performs spectral feature comparison and reflectivity determination on the real set of changed pixels, subdividing the hidden dangers into high reflectivity changes and surface darkening; finally, it associates the attribute table of the changed patches with the equipment ledger space to generate a list of detected hidden danger changes. This invention effectively reduces the false alarm rate, clarifies the physical attributes of hidden dangers, and enables rapid location of the changed area to specific equipment, improving the accuracy and practicality of substation hidden danger monitoring. Attached Figure Description

[0022] 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.

[0023] Figure 1 This is a schematic diagram of the overall process of a substation hazard change detection method according to an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram illustrating the process and determination of the actual change pixel set in the substation hidden danger change detection method according to an embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram illustrating the process of summarizing the set of real high reflectivity change pixels in the substation hidden danger change detection method according to an embodiment of the present invention. Detailed Implementation

[0026] 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.

[0027] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for detecting changes in potential hazards in a substation is provided, comprising: S1, in response to the calculation of the solar vector in the second time phase, constructs a simulated shadow map; S2 acquires the original change map of the second time phase compared with the first time phase, and performs a first-level filtering mechanism on the simulated shadow map and the original change map. The first-level filtering mechanism is used to filter out the set of real change pixels. S3, compare the actual set of changed pixels with the set of spectral feature vectors collected in the first time phase to obtain the actual set of highly reflective changed pixels; S4. The original change map is stitched together with the real set of high reflectivity change pixels to generate a hazard attribute classification map. S5, perform adjacent spatial clustering on the hazard attribute classification map to generate change patches, and obtain the change patch attribute table through the parameters of the change patches; S6. Associate the changed patch attribute table with the equipment parameters of the substation to establish a hidden danger change detection list.

[0028] The first time phase refers to satellite imagery acquired at an earlier time point, serving as a baseline or reference image for change detection. The second time phase refers to satellite imagery acquired at a later time point, serving as a detection image or target image for change detection.

[0029] This scheme identifies changes in land cover or equipment status by comparing images of the same area acquired at two different time phases, namely the first time phase and the second time phase.

[0030] Existing technologies utilize optical imagery from multi-source satellite data combined with synthetic aperture radar for substation hazard detection. However, this approach suffers from three major drawbacks: First, it directly performs pixel-by-pixel differential comparison, failing to distinguish between equipment shadow movement and actual surface changes, resulting in a high false alarm rate. Second, the change analysis is limited to spectral or texture differences, only indicating brightening or darkening without clear physical meaning, and cannot be directly linked to specific risks. Third, the detection results are disconnected from equipment records, failing to provide clear guidance for hazard investigation and leading to low efficiency in handling the situation.

[0031] This invention calculates the solar vector and constructs a simulated shadow map, eliminating negative brightness pixels falling into the shadow area of ​​the simulated shadow map to ensure the authenticity of the detection. Then, it compares the spectral features and determines the reflectivity of the actual set of changed pixels, subdividing potential hazards into high reflectivity changes and surface darkening. Finally, it links the attribute table of the changed patches with the equipment ledger space to generate a list of potential hazard changes. This invention effectively reduces the false alarm rate, clarifies the physical attributes of potential hazards, and enables rapid location of changed areas to specific equipment, improving the accuracy and practicality of substation hazard monitoring.

[0032] Example 2, refer to Figure 1 As an embodiment of the present invention, based on the above embodiment, a method for detecting changes in hidden dangers in substations is provided, comprising the following steps: S1. Calculate the solar vector in response to the second time phase and construct a simulated shadow map; The steps for constructing the simulated shadow map include S1.1~1.4: S1.1 Obtain the image information features of the second time phase and retrieve the geographic coordinates of the device under test; The computer reads the imaging timestamp from the image information features of the second phase and simultaneously retrieves the satellite orbital parameters and attitude data, with a roll angle of 0.1° and a pitch angle of 0.2°.

[0033] This data shows that the satellite exhibits a tilt of 0.1° around its front and rear axes and a pitch deviation of 0.2° around its left and right axes during flight. Although the angles are small, at an altitude of several hundred kilometers, they can cause a nearly 2-kilometer shift in the field of view for ground observation. Therefore, this type of data is a key input parameter for subsequent geogeometric correction of remote sensing imagery.

[0034] Next, the WGS-84 geographic coordinates of the equipment under test were retrieved from the substation equipment ledger database. The longitude is 116.40°E and the latitude is [missing information]. The value is 39.90°N. These data collectively form the foundation for subsequent calculations of the sun's position.

[0035] S1.2, Perform function calculations on the image information features of the second time phase and the geographic coordinates of the device under test to obtain the solar position parameters relative to the device under test; First, we read the image information features of the second time phase and obtained the solar declination δ = -14.2° and the solar hour angle H = 40.5°.

[0036] Calling the solar declination Solar hour angle and latitude Perform spherical trigonometric function operations: ; Substituting the data δ = -14.2°, H = 40.5°, =39.90°, which can be calculated ≈0.581, h≈35.5°.

[0037] Continuing with the above data, perform the solar azimuth calculation using the azimuth formula. ; The azimuth angle A = 210.2° is calculated using the above formula. All the parameters obtained above are the solar position parameters corresponding to the device under test, which can be used for subsequent process determination calculations.

[0038] S1.3, Simulate the projection relationship between light and shadow based on the solar position parameters to obtain shadow distribution data; The relationship between light and shadow projection is essentially about determining whether the light emitted from the sun and shining on each grid point on the ground is blocked by the device under test. If it is blocked, the grid point is a shadow area; otherwise, it is a lit area.

[0039] To determine whether a grid point on the ground is obscured by the device under test (DUT), we must first know the shape, height, and spatial location of the DUT. Therefore, the computer retrieves the three-dimensional geometric models of the DUT within the substation. These models consist of a series of triangular facets that precisely describe the outer surface contour of each DUT.

[0040] One device under test can be selected as the center, and a 10m × 10m area on the ground around it can be designated as the sampling zone, with a sampling interval of 0.1m. At the same time, the 3D model data of other devices under test within the area can be loaded.

[0041] For each grid point within the sampling area, a virtual ray is emitted from that grid point in the opposite direction of sunlight, and it is checked whether the ray intersects with any loaded device model surface. If the ray intersects with any triangular facet of the 3D model, it means that the grid point is blocked by the corresponding device under test and cannot receive sunlight, thus being identified as a shadow area; conversely, if the ray does not intersect with any 3D model, it is identified as an illuminated area.

[0042] The following is a more detailed analysis, using the following formula to convert the sun's position parameters into a normalized sun vector in the scene coordinate system: ; Starting from each grid point P(x, y, z) in the sampling area, along... A ray is emitted in a specific direction. The system checks if the ray intersects with any of the triangular faces of the 3D model. If the ray does not intersect with any other triangular facet, P(x, y, z) is considered an illuminated area, and the light intensity is calculated. If the ray intersects with any other triangular facet, P(x, y, z) is considered a shadow area, and the light intensity is 0.

[0043] After determining all grid points, we can obtain the shadow distribution data of the device under test.

[0044] S1.4, Render the shadow distribution data to generate a simulated shadow map.

[0045] Combining the grid points from the above steps And the set of light intensity, and then substituting the sunlight incident angle parameters of the second phase, h≈35.5°, A=210.2°.

[0046] The computer sets up a virtual camera with the same viewing angle as the satellite in the second time phase, using the same roll angle of 0.1° and pitch angle of 0.2° as the satellite, and defines the output image resolution as 1000×1000 pixels.

[0047] Create a two-dimensional image matrix and then use the grid points in the three-dimensional sampling area. Projected onto a two-dimensional image matrix, for each pixel (i, j), a value is assigned based on the illumination intensity of the corresponding spatial point. If the pixel (i, j) corresponds to a three-dimensional mesh point... If the light intensity is 0, then pixel (i, j) is assigned the shadow grayscale value RGB(50,50,50), and is located in the shadow area.

[0048] If the light intensity is greater than 0, assign the value RGB(150,150,150) to pixel (i,j) and place it in the illuminated area. After all pixels (i,j) have been assigned values, the simulated shadow map for the second phase can be generated.

[0049] S2, obtain the original change map of the second time phase compared with the first time phase, and perform a first-level screening mechanism on the simulated shadow map and the original change map. The first-level screening mechanism is used to screen out the set of real change pixels.

[0050] refer to Figure 2 As shown, Figure 2 The process and determination details for obtaining the set of real changed pixels in step S2 are given.

[0051] S2.1, in response to the extraction of image information features from the first time phase and image information features from the second time phase, establish a feature mapping relationship; The computer reads geographic coordinates from the image information features of the first time phase and the second time phase, respectively, with each image corresponding to the actual location. Because there may be slight differences in the satellite's attitude and orbit between the two shooting times, the geographic coordinates corresponding to the same location in the image information features of the first and second time phases will not be exactly the same.

[0052] Therefore, the computer needs to establish a feature mapping relationship to describe the rules between the geographic coordinates of a location in the second time phase and the geographic coordinates of the corresponding location in the first time phase.

[0053] S2.2, Align the image information features of the first time phase and the image information features of the second time phase according to the feature mapping relationship to obtain an intermediate map; After obtaining the feature mapping relationship, the computer adjusts the image information features of the second time phase based on this relationship. For a position within the image information features of the second time phase, the feature mapping relationship determines where that position should be located within the image information features of the first time phase. A corresponding position grid is established within the image information features of the first time phase to facilitate matching with the image information features of the second time phase.

[0054] Next, the computer resamples the positions in the image information features of the second time phase into the position grid inside the image information features of the first time phase. After this step, the positions corresponding to the same geographic coordinates in the image information features of the second time phase and the image information features of the first time phase coincide, thus forming the intermediate map we need. Subsequent calculations can be performed directly based on the intermediate map.

[0055] S2.3 Calculate the brightness difference value at each position using the intermediate spectrum, and generate the original variation spectrum based on all the brightness difference values.

[0056] After the overlap is completed, the computer queries each location and reads the brightness value B of the first phase image information feature at that location. T1 And in the intermediate atlas, the brightness value B at that location in the image information features of the second phase. T2 Through B T2 -B T1 Calculate the difference between the two to obtain the brightness difference value at this location.

[0057] In this embodiment, a location with coordinates (10, 20) is selected, and the brightness value B at this location is... T1 150, B T2 The brightness value is 80, and the resulting brightness difference is 80-150=-70. Since -70 < 0, it means that the position (10,20) has become darker. If the brightness difference value at this position is greater than 0, it means that the position has become brighter. The computer uses this logic to determine all positions and saves them as a new map, which is the original change map. The original change map initially marks all areas where brightness and darkness changes may occur, but it has not yet eliminated false changes caused by shadow movement, so further steps are needed for filtering.

[0058] The steps for performing a primary screening mechanism on the simulated shaded map and the original change map include S2.4~S2.8: S2.4, pre-identify shaded and non-shaded areas within the simulated shadow map; In step S1, the computer has already generated a simulated shadow map. The simulated shadow map is an image rendered based on the second-phase solar position parameters and the 3D model of the device under test. In the simulated shadow map, each location is explicitly marked as one of two states: if the location is blocked from receiving sunlight, it is marked as a shadow area; if sunlight can directly reach the location, it is marked as a non-shadow area. In this embodiment, the computer assigns a value of 0 to shadow areas and a value of 1 to non-shadow areas. The computer directly reads this marking information corresponding to the simulated shadow map without requiring additional processing or modification of the simulated shadow map. The results of this pre-identification will be used for judgment in subsequent steps.

[0059] S2.5 In the original change map, the positions with a brightness difference value less than 0 are taken as negative pixels, and the positions with a brightness difference value greater than 0 are taken as positive pixels, generating a set of negative pixels and a set of positive pixels; The computer opens the original change map generated in S2.3. Each location in the original change map corresponds to a brightness difference value. The computer reads the brightness difference value of location (10,20) in S2.1 as -70. Therefore, the brightness difference value of this location is less than 0, indicating that this location is darker in the second time phase compared to the first time phase. However, it should be noted that this darkening may be due to actual darkening of the ground or a pseudo-change caused by shadow movement.

[0060] Locations with a brightness difference value less than 0 are recorded in the negative pixel set, and locations with a brightness difference value greater than 0 are recorded in the positive pixel set. After the computer finishes reading, it obtains the negative pixel set and the positive pixel set.

[0061] S2.6 Query the position corresponding to each position in the simulated shadow map in the negative pixel set, and take the obtained position as the feature position; Since shadows can only darken the ground and not lighten it, only the shadow fidelity needs to be verified on the negative pixel set. The computer retrieves the coordinates of each location in the negative pixel set one by one, and then searches for them in the simulated shadow map identified in S3.1. The location corresponding to these coordinates in the shadow map is called the eigenlocation. The eigenlocation is essentially the state of the same ground point in the shadow map, telling us whether the eigenlocation is covered by shadow in the theoretical simulation involved in the simulated shadow map, so as to make subsequent judgments.

[0062] S2.7 When a feature is located in a shaded area, it is removed from the negative pixel set; when a feature is located in a non-shaded area, it is retained in the negative pixel set. The computer checks whether each feature location belongs to a shaded or unshaded area in the simulated shadow map. Using the location (10,20) again, if the value for the feature location (10,20) in the simulated shadow map is 0, it means that the feature location (10,20) is marked as a shaded area, and (10,20) is indeed covered by the shadow of the device under test. This location appears as darkened in the original change map, which is a false change caused by shadow movement, rather than a true change in ground physics. Therefore, the computer removes (10,20) and locations of this type from the set of negative pixels.

[0063] Conversely, if the trait (10,20) is assigned a value of 1, it is marked as a non-shaded area, meaning that (10,20) itself is not within the shadow of any device. Therefore, its darkening must be a real ground change, such as land subsidence, oil spills, or vegetation withering, and should be preserved. After the computer has checked all negative pixels and completed the removal process, the remaining set of negative pixels is the true set of negative pixels, representing the actual darkening changes that have occurred.

[0064] S2.8 The negative pixel set and the positive pixel set are merged to form the real change pixel set.

[0065] The computer merges the set of true negative pixels retained from the previous step with the initial set of positive pixels. The positive pixel set does not need to remove shadows because shadows do not cause brightening. Therefore, this step yields a complete set of true change pixels, containing all locations where changes have actually occurred. Positive pixels represent brightened areas, and true negative pixels represent darkened areas.

[0066] In another alternative implementation, the primary screening mechanism can also employ other existing image change detection and screening methods. For example, a pixel-threshold-based segmentation method can be used: setting a brightness difference threshold of -20, directly identifying locations with brightness differences below -20 as candidate locations, and then eliminating them in conjunction with a simulated shadow map. Alternatively, texture feature-based change detection can be used, comparing the local texture statistics of the first and second phase influence information features, identifying texture abrupt change regions as candidate locations, and then eliminating locations within the shadow area of ​​the simulated shadow map.

[0067] Alternatively, screening can be based on changes in vegetation indices. By using the normalized vegetation index (NDI) of the first and second time phases to identify the influence information features, areas with a significant decrease in the NDI can be marked as candidate locations. Then, simulated shadow maps can be used to eliminate spurious changes in shadows.

[0068] S3, compare the actual set of changed pixels with the set of spectral feature vectors collected in the first time phase to obtain the actual set of highly reflective changed pixels; The steps to obtain the true set of highly reflective variation pixels include S3.1~3.6.

[0069] refer to Figure 3 As shown, Figure 3 The process and determination details for obtaining the set of real high reflectivity variation pixels in step S3 are presented.

[0070] S3.1, the image information features of the first time phase are divided into a preset background category set, and the first average spectral feature vector corresponding to each category in the background category set is calculated; The computer first pre-defines a set of background categories, such as five categories including vegetation, bare soil, buildings, water bodies, and equipment areas. These categories cover the most common stable land cover types in and around the substation. Then, it acquires the spectral information of each location in the first temporal image information features, extracts the reflectance values ​​of the blue, green, red, and near-infrared bands, and assigns each location in the first temporal image information features to the aforementioned pre-defined background category. After the classification is completed, the first average spectral feature vector is calculated using the following formula; .

[0071] S3.2, combine all the first average spectral eigenvectors into a first spectral eigenvector set; Obtain the first average spectral feature vectors 1, 2, 3, 4, and 5 calculated in step S4.1. Combine these first average spectral feature vectors together to obtain the first spectral feature vector set {1, 2, 3, 4, 5}.

[0072] S3.3, Based on the real change pixel set, a two-level filtering mechanism is executed to divide the result of the two-level filtering mechanism into a bright pixel set and a dark pixel set; The computer retrieves the set of real change pixels obtained from S3.5, where each location is accompanied by the brightness difference value from the original change map. A secondary filtering mechanism sets a threshold of ±10, with an actual value of 10. A value of 10 is significantly greater than 0, used to distinguish between significant and subtle changes.

[0073] The computer queries the data in the set of all real changes in pixels. For example, the brightness difference value of position (50,50) is 15. Since 15 is greater than 10, position (50,50) is judged to be a significant change and is included in the set of brightened pixels.

[0074] The brightness difference value of the example position (10,20) in S2.3 is -70. Since -70 is less than -10, position (10,20) is judged to be significantly darkened and is included in the set of darkened pixels.

[0075] If the brightness difference value is between -10 and 10, it is considered that the change is not significant and will not be included in the subsequent detailed analysis. Through this step, the locations corresponding to the set of pixels with real changes are further subdivided into bright areas that need to be analyzed in detail and dark areas that can be directly classified.

[0076] S3.4, Obtain the corresponding brightness spectral features of each position in the brightness pixel set in the image information features of the second time phase, and generate a brightness spectral feature set; For each location in the set of illuminated pixels obtained in S3.3, the computer reads the illumination spectral characteristics at that location from the second temporal image information features, i.e., the reflectance values ​​of each blue, green, and red band. We still choose location (50, 50), and the reflectance value of the corresponding band is (R... B =0.5,R G =0.6,R R =0.7).

[0077] Since the second phase image information features have already been geometrically aligned with the first phase image information features in S2.2, their positions can be directly matched without additional registration.

[0078] The brightness spectral features of each location are collected together to form a brightness spectral feature set. Each data point in the brightness spectral feature set represents the actual spectral curve of a significantly brightened location in the second time phase, which is subsequently used for comparison with the background spectrum.

[0079] S3.5, calculate and compare the set of brightened spectral features with the set of first spectral feature vectors to obtain the comparison value; Still referencing position (50, 50), calculate the average reflectance of the three bands at position (50, 50), which is (0.5 + 0.6 + 0.7) / 3 = 0.6. The average reflectance at position (50, 50) is 0.6. At this point, we set a high reflectance reference value of 0.45. The high reflectance reference value of 0.45 is calculated based on the information characteristics of the first time image. The calculation method is the same as the above method, which calculates the average reflectance. The minimum value of the final average reflectance obtained in the first time image is 0.45. As a reference value, it is set as the high reflectance reference value of 0.45 in this step for comparison with the average reflectance in the brightness spectral feature set later.

[0080] S3.6, Based on the comparison value, a three-level screening mechanism is executed to generate a set of real high reflectivity change pixels corresponding to the positions of the three-level screening mechanism results; The average reflectance of position (50, 50), which is 0.6, is compared with the high reflectance benchmark value of 0.45. Since 0.6 is greater than 0.45, the average reflectance exceeds the high reflectance benchmark value. Therefore, position (50, 50) is included in the set of true high reflectance variation pixels. Other positions that meet the above criteria are also included in the set of true high reflectance variation pixels. If the above criteria are not met, the position is not included in the set of true high reflectance variation pixels.

[0081] S4. The original change map is stitched together with the real set of high reflectivity change pixels to generate a hazard attribute classification map. The steps for stitching the original variation map with the real set of high reflectivity variation pixels to generate a hazard attribute classification map include S4.1~4.4.

[0082] S4.1, Create a blank matrix by importing all positions corresponding to the real high reflectivity change pixel set and the darkening pixel set into the blank matrix to generate a preliminary parameter matrix; The computer first creates a blank matrix with the same dimensions as the original image, with each element in the matrix initialized to 0.

[0083] Subsequently, the coordinates of each position are extracted from the set of true high-reflectivity change pixels obtained in step S4.6, and these coordinate positions are recorded in a blank matrix. Similarly, the coordinates of each position are extracted from the set of darkened pixels obtained in S4.3, and these coordinate positions are also recorded in the same blank matrix. At this point, the matrix only marks which positions belong to the two types of change pixels, but has not yet been assigned category discrimination values. We call this current matrix the preliminary parameter matrix.

[0084] S4.2 Assign first category values ​​to all positions corresponding to the set of real high reflectivity variation pixels in the preliminary parameter matrix; The computer queries the preliminary parameter matrix for all positions belonging to the set of true high-reflectivity variation pixels and modifies the values ​​of the corresponding elements in the matrix from 0 to the first category value. The first category value is used to identify the type of true high-reflectivity variation, for example, it is set to the integer 1. After the assignment is completed, these positions are no longer 0 in the matrix, but have a clear category identifier.

[0085] S4.3, assign the second category value to all positions corresponding to the set of darkened pixels in the preliminary parameter matrix; Similarly, the computer queries the preliminary parameter matrix for all positions belonging to the darkened pixel set and modifies the value of the corresponding element in the matrix from 0 to a second category value. The second category value is used to identify the surface darkening type, for example, set to an integer 2. The positions in the darkened pixel set, which were originally imported in S5.1, are now assigned a category value different from the actual high reflectivity changes for subsequent differentiation.

[0086] S4.4, based on the preliminary parameter matrix that has been assigned values ​​for the first category and the second category, generate a hazard attribute classification map.

[0087] The computer directly outputs the preliminary parameter matrix after the above assignment as a hazard attribute classification map. In the map, pixels with a value of 1 represent locations of actual high reflectivity changes, such as illegal metal structures or scattered foil pieces within the area; pixels with a value of 2 represent darkened locations, such as withered vegetation or oil spills within the area; and pixels with a value of 0 represent background areas where no actual changes have occurred. At this point, the hazard attribute classification map is complete, and it will be used in subsequent spatial clustering and patch extraction steps.

[0088] S5, perform adjacent spatial clustering on the hazard attribute classification map to generate change patches, and obtain the change patch attribute table through the parameters of the change patches; The steps for obtaining the attribute table of the changed patch through the parameters of the changed patch include S5.1~5.5.

[0089] S5.1 Identify adjacent locations with the same category value in the hazard attribute classification map, and merge the identified locations to generate change patches; The computer reads the hazard attribute classification map. The value of each pixel in the hazard attribute classification map includes the first category value 1, which corresponds to the actual high reflectivity change position, the second category value 2, which corresponds to the darkening position, and 0.

[0090] Next, the computer uses spatial adjacency rules to traverse all non-zero pixels in the graph, find positions within 8 neighborhoods, check if the category values ​​of the positions within the 8 neighborhoods are the same, and if they are the same, merge all these positions into the same connected region. Repeat this process until all positions with the same category value in 8 neighborhoods have been merged.

[0091] Each merged connected region is a change patch. Ultimately, we obtain the change patches corresponding to the real high reflectivity changes and the change patches corresponding to the darkening changes.

[0092] S5.2 Extract the coordinate data corresponding to the position in each changed patch, calculate the average coordinate value, and use it as the geometric center coordinate of each changed patch.

[0093] For each changed feature, the computer extracts the coordinate data of all locations within that feature. For example, if there are 10 locations in a changed feature, then the coordinate data of 10 such locations is extracted. The set of location coordinate data is as follows: .

[0094] Summing the x-coordinates of all positions using set operations. Summing the y-coordinates of all positions using set operations. The summation result is divided by the total number of positions. Then we can obtain the coordinates of the geometric center. .

[0095] S5.3 Obtain the spatial resolution from the image information features of the first or second time phase, and calculate the area of ​​each changed patch based on the obtained spatial resolution and the number of positions in each changed patch.

[0096] The computer extracts the spatial resolution parameter R from the image information features of the first or second time phase. The spatial resolution parameter R represents the actual ground length corresponding to each location; for example, R = 0.5 meters per location. For each changed patch, the total number of locations K is counted, and then the total number of locations K is multiplied by the square of the spatial resolution R. 2 R 2 This refers to the ground area at a single location, which gives the actual area of ​​the changed patch.

[0097] Taking a total number of locations K=10 as an example, the actual area of ​​the changed patch is 10 * 0.25 = 2.5 square meters.

[0098] S5.4 Extract the elevation data corresponding to the location in each changed patch, calculate the average elevation, and obtain the average elevation of each changed patch.

[0099] The computer directly retrieves the elevation data of each changed patch. Elevation data The surface elevation of each location was recorded.

[0100] For example The range is 50.0~50.8 meters. At this point, the average elevation is calculated, and the elevation data for all locations are... Summation of set operations The summation result is divided by the total number of positions. ,get mean elevation ,For example Meters. Following this process, repeat the calculation for all remaining changed features to obtain the average elevation of each changed feature.

[0101] S5.5 generates a change patch attribute table in response to the sum of the geometric center coordinates, area, and average elevation of each change patch.

[0102] The computer summarizes the geometric center coordinates, area, and average elevation of each changed feature into a structured table, resulting in the changed feature attribute table. This attribute table provides the foundational data for subsequent spatial association with the equipment ledger.

[0103] S6. Associate the changed patch attribute table with the equipment parameters of the substation to establish a hidden danger change detection list.

[0104] The computer reads the logbook information of each major piece of equipment from the substation's equipment management database. This equipment includes transformers, combined electrical appliances, surge arresters, etc. Each logbook record contains the equipment number, such as equipment A, and the geographical coordinates of equipment A, longitude 116.401°N, latitude 39.902°N.

[0105] Convert the geographic coordinates of device A to the corresponding planar coordinates in the projected coordinate system under the same image. For example, the planar coordinates of device A are (448500.1, 4417050.2), which is convenient for subsequent distance calculation.

[0106] The computer retrieves the attribute table of the changed features generated by S6.5, which records the geometric center coordinates (448501.0, 4417051.5) of each changed feature. Using the geometric center coordinates and the planar coordinates of device A, the planar distance between the geometric center and device A is calculated, yielding the desired result. rice.

[0107] ; Following the same calculation logic, we calculate the planar distances between other devices and the geometric center, and finally select the minimum planar distance. Here, we assume the minimum planar distance... = rice.

[0108] For each changed patch, the computer will merge the nearest device number and minimum distance with the area, average elevation, and category values ​​1 and 2 in the hazard attribute classification map that are already in the changed patch attribute table.

[0109] Record this information one by one according to the different change patches to generate a list table. This table is the hazard change detection list. Maintenance personnel can directly locate specific equipment and hazard types based on the list without manual interpretation.

[0110] Example 3 illustrates a schematic scheme for a substation hazard change detection method. It should be noted that the technical solution of this substation hazard change detection system belongs to the same concept as the technical solution of the aforementioned substation hazard change detection method. Details not described in detail in this embodiment can be found in the description of the technical solution of the aforementioned substation hazard change detection method.

[0111] This embodiment also provides a substation hazard change detection system, including: The simulated shadow construction module is used to construct a simulated shadow map in response to the calculation of the solar vector in the second time phase; The original change map acquisition module is used to acquire the original change map of the second time phase compared with the first time phase; The first-level filtering module is used to perform a first-level filtering mechanism on the simulated shadow map and the original change map, and uses the result of the first-level filtering mechanism as the set of real change pixels. The spectral comparison module is used to compare the true set of changed pixels with the spectral feature vector set of the first time phase to obtain the true set of highly reflective changed pixels. The classification map generation module is used to stitch the original change map with the real set of high reflectivity change pixels to generate a hazard attribute classification map; The image clustering module is used to perform adjacent spatial clustering on the hidden danger attribute classification map, generate change images, and obtain the change image attribute table through the parameters of the change images. The list creation module is used to associate the changed patch attribute table with the equipment parameters of the substation to create a list of potential hazard changes.

[0112] This embodiment also provides an electronic device suitable for detecting changes in substation hazards, 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 implement the substation hazard change detection method proposed in the above embodiment.

[0113] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the substation hazard change detection method proposed in the above embodiments.

[0114] The storage medium proposed in this embodiment and the method for detecting changes in substation hazards 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.

[0115] 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.

[0116] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended 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.

Claims

1. A method for detecting changes in potential hazards in a substation, characterized in that, include: In response to the calculation of the solar vector in the second time phase, a simulated shadow map is constructed; Obtain the original change map of the second time phase compared with the first time phase, and perform a first-level filtering mechanism on the simulated shadow map and the original change map. The first-level filtering mechanism is used to filter out the set of real change pixels. The true set of changed pixels is compared with the set of spectral feature vectors collected in the first time phase to obtain the true set of highly reflective changed pixels. The original variation map is stitched together with the real set of high reflectivity variation pixels to generate a hazard attribute classification map; The hazard attribute classification map is clustered by adjacent spaces to generate change patches, and the change patch attribute table is obtained through the parameters of the change patches; By associating the attribute table of the changed patches with the equipment parameters of the substation, a list of potential hazard changes can be established.

2. The method for detecting changes in substation hazards as described in claim 1, characterized in that, The steps to construct a simulated shadow map include: Acquire the image information features of the second time phase and retrieve the geographic coordinates of the device under test; The image information features of the second time phase and the geographic coordinates of the device under test are used to perform function calculations to obtain the solar position parameters relative to the device under test; The shadow distribution data is obtained by simulating the relationship between illumination and shadow projection based on the solar position parameters. The shadow distribution data is rendered to generate a simulated shadow map.

3. The method for detecting changes in substation hazards as described in claim 2, characterized in that, The steps to obtain the original change spectrum of the second time phase compared to the first time phase include: In response to the extraction of image information features from the first time phase and image information features from the second time phase, a feature mapping relationship is established; The image information features of the first time phase and the image information features of the second time phase are aligned according to the feature mapping relationship to obtain an intermediate map. The brightness difference value at each position is calculated using the intermediate spectrum, and the original variation spectrum is generated based on all the brightness difference values.

4. The method for detecting changes in substation hazards as described in claim 3, characterized in that, The steps for performing a primary screening mechanism on the simulated shaded map and the original change map include: Pre-identify shaded and non-shaded areas within the simulated shadow map; The positions with a brightness difference value less than 0 in the original change map are taken as negative pixels, and the positions with a brightness difference value greater than 0 are taken as positive pixels, thus generating a set of negative pixels and a set of positive pixels. In the set of negative pixels, query the position corresponding to each position in the simulated shadow map, and use the obtained position as the feature position; If a feature is located in a shaded area, it is removed from the negative pixel set; if a feature is located in a non-shaded area, it is retained in the negative pixel set. The negative pixel set and the positive pixel set are merged to form the true change pixel set.

5. The method for detecting changes in substation hazards as described in claim 1, characterized in that, The steps for comparing the actual set of changed pixels with the set of spectral feature vectors acquired in the first time phase include: The image information features of the first time phase are divided into a preset background category set, and the first average spectral feature vector corresponding to each category in the background category set is calculated; Combine all the first average spectral eigenvectors into a first spectral eigenvector set; A two-level filtering mechanism is executed based on the real change pixel set, and the result of the two-level filtering mechanism is divided into a brightened pixel set and a darkened pixel set. Obtain the corresponding brightness spectral features of each position in the brightness pixel set in the image information features of the second time phase, and generate a brightness spectral feature set; The set of brightened spectral features is compared with the set of first spectral feature vectors to obtain the comparison value; A three-level screening mechanism is executed based on the comparison values, and a set of real high reflectivity variation pixels is generated at the positions corresponding to the results of the three-level screening mechanism.

6. The method for detecting changes in substation hazards as described in claim 5, characterized in that, The steps for stitching the original variation map with the actual set of highly reflective variation pixels include: Create a blank matrix and import all positions corresponding to the real high reflectivity change pixel set and the darkening pixel set into the blank matrix to generate a preliminary parameter matrix; Assign first category values ​​to all positions corresponding to the set of real high reflectivity variation pixels in the preliminary parameter matrix; Assign second category values ​​to all positions corresponding to the set of darkened pixels in the initial parameter matrix; Based on the preliminary parameter matrix that has been assigned values ​​for the first and second categories, a hazard attribute classification map is generated.

7. The substation hazard change detection method as described in claim 6, characterized in that, The steps of generating change patches and obtaining a change patch attribute table based on the parameters of the change patches include: Identify adjacent locations with the same category value in the hazard attribute classification map, and merge the identified locations to generate change patches; Extract the coordinate data corresponding to the location of each changed patch, and calculate the mean of the coordinates. As the coordinates of the geometric center of each changing patch; Spatial resolution is obtained from the image information features of the first or second time phase. Based on the obtained spatial resolution and the number of locations in each changed patch, the area of ​​each changed patch is calculated. Extract the elevation data corresponding to the location in each changed patch, calculate the average elevation, and obtain the average elevation of each changed patch; In response to the sum of the geometric center coordinates, area, and average elevation of each changed feature, an attribute table of the changed feature is generated.

8. A substation hazard change detection system, employing the substation hazard change detection method as described in claim 7, characterized in that, include: The simulated shadow construction module is used to construct a simulated shadow map in response to the calculation of the solar vector in the second time phase; The original change map acquisition module is used to acquire the original change map of the second time phase compared with the first time phase; The first-level filtering module is used to perform a first-level filtering mechanism on the simulated shadow map and the original change map, and uses the result of the first-level filtering mechanism as the set of real change pixels. The spectral comparison module is used to compare the true set of changed pixels with the spectral feature vector set of the first time phase to obtain the true set of highly reflective changed pixels. The classification map generation module is used to stitch the original change map with the real set of high reflectivity change pixels to generate a hazard attribute classification map; The image clustering module is used to perform adjacent spatial clustering on the hidden danger attribute classification map, generate change images, and obtain the change image attribute table through the parameters of the change images. The list creation module is used to associate the changed patch attribute table with the equipment parameters of the substation to create a list of potential hazard changes.

9. An electronic device, characterized in that, include: 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 hidden danger change detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of changing substation hazards as described in any one of claims 1 to 7.

Images