Method for evaluating the operating state and aging mechanism characteristics of GIS pot-type insulators

Abstract

The application relates to the technical field of insulation monitoring, in particular to an evaluation method for the operation state and aging mechanism characteristics of a GIS basin-type insulator, which comprises the following steps: calling image and material information, analyzing grain morphology, screening crack fragments, aggregating to generate a distribution diagram, extracting a thermal diffusion and stress diagram, forming a stress conduction diagram, judging a breakdown path overlap, generating an electrical path diagram, dividing an aging block, classifying a synchronous area, and outputting a trend and mechanism conclusion. The application extracts grain arrangement and morphology mutation characteristics, constructs a structure aggregation layer, combines a thermal diffusion and stress trend relationship, generates a stress conduction spectrum, identifies an overlapping section of a current and a discharge track, marks a track repetition and a turning dense area, superimposes a peeling and crack boundary, divides an aging block, extracts synchronous damage and stage aging behavior, and supports state evaluation and path analysis.

Description

Technical Field

[0001] This invention relates to the field of insulation monitoring technology, and in particular to a method for evaluating the operating status and aging mechanism characteristics of GIS pot-type insulators. Background Technology

[0002] The field of insulation monitoring technology encompasses techniques for monitoring and evaluating the condition of insulation systems in power equipment. Its core content mainly includes real-time monitoring and analysis of the performance, aging, and potential faults of various types of insulation materials and equipment during operation. Insulation monitoring technology detects changes in the condition of components such as insulators, conductors, and switches in power equipment, enabling timely detection of insulation faults or potential hazards to ensure the safe operation of the power system. This technology is widely used in high-voltage power equipment, substations, distribution networks, and other power facilities, covering methods for detecting and diagnosing problems such as surface contamination, partial discharge, and aging of insulators. Real-time monitoring of the insulator's operating status can effectively extend equipment lifespan and improve the reliability and safety of the power system.

[0003] The evaluation method for the operating status and aging mechanism characteristics of GIS basin insulators refers to a technical solution for assessing the operating status and aging characteristics of basin insulators in GIS equipment. It mainly involves establishing an evaluation method by monitoring the changes in the state of basin insulators during long-term operation and analyzing different aging mechanisms. Specifically, it involves real-time monitoring of physical quantities such as vibration, temperature, and electric field of the insulators, combined with aging characteristic models, to analyze the possible aging processes and failure risks during operation. This includes identifying factors such as surface contamination, partial discharge, and material aging of the insulators, and through multi-dimensional data collection and analysis, quantitatively assessing the health status of the insulators, thereby providing a basis for the operation and maintenance of power equipment.

[0004] Existing technologies have gaps in identifying the evolution characteristics of micro-regions in structures. In particular, it is difficult to accurately extract grain reconstruction, crack intersections, and stress concentration areas in the edge structure of ceramic bodies. There is a lack of a way to establish a multi-physical quantity time-series fusion path, which makes it impossible to accurately restore the spatial coupling characteristics of current trajectories and discharge channels. When discontinuous damage occurs on the surface of the material and the operating path overlaps, problems such as blurred region division and misjudgment of failure mechanisms often occur, affecting the completeness of aging mechanism identification and the timeliness of operational risk response. Summary of the Invention

[0005] To address the technical problems existing in the prior art, embodiments of the present invention provide a method for evaluating the operating status and aging mechanism characteristics of GIS basin insulators. The technical solution is as follows:

[0006] The evaluation method for the operating status and aging mechanism characteristics of GIS pot insulators includes the following steps:

[0007] Images and material responses of the edge ceramic body and conductive contact area of ​​the GIS basin insulator are retrieved, the grain arrangement and continuity of the connecting band are analyzed, the structure of lattice interruption and crack intersection is screened, the boundary abrupt change trend is compared, different orientation markers are superimposed, spatial clustering is performed, and a structural evolution distribution map is generated.

[0008] Based on the distribution location of structural mutations in the structural evolution distribution map, the time series layers of the heat diffusion map and stress migration map are extracted, the intersection relationship between stress trajectory and structural mutation zone is analyzed, the overlapping segments of stress guidance and electric field line direction are screened, the intersection points are recorded, and a multi-source stress transmission direction map is generated.

[0009] By calling the densely turning sections in the multi-source stress transmission path diagram, comparing the current mutation trajectory diagram with the discharge channel diagram, analyzing the spatial overlap of multiple breakdown directions, filtering repeated intersection trajectories, marking the path direction distribution characteristics, and generating an electrical response transfer path diagram;

[0010] Based on the intersection area of ​​the electrical response transfer path diagram, the corresponding material peeling diagram and crack connectivity diagram are retrieved to determine the range of surface color difference superposition and crack edge abrupt change, the concentrated area of ​​physical damage is screened, the layer boundary is unified, and the structurally consistent blocks are divided to form an aging behavior mapping group. The operating status and aging mechanism characteristics of the insulator are evaluated by analyzing the characteristics of the aging behavior mapping group.

[0011] As a further aspect of the present invention, the structural evolution distribution map includes grain arrangement direction change partitions, boundary morphology abrupt change type identification, and lattice interruption and crack intersection location; the multi-source stress transmission path map includes thermal-stress coupling path distribution, effective stress transmission nodes, and stress and electric field direction consistent sections; the electrical response transfer path map includes current trajectory overlap distribution, discharge channel intersection characteristics, and spatial path dense sections; and the aging behavior mapping group includes surface peeling concentrated patches, crack abrupt change boundary combinations, and color difference superposition region sets.

[0012] As a further aspect of the present invention, the process of generating the structural evolution distribution map is as follows:

[0013] Images and material response information of the ceramic body area and conductive contact area at the edge of the GIS basin insulator are retrieved. The grain orientation information in the image is vectorized, and various response data related to the grain orientation are extracted from the material response information. Based on the continuity of the grain orientation vector change in the regional image, the orientation state of the connecting strip edge is analyzed. At the same time, combined with the contact distribution state of the adjacent area edge, the boundary segment where the orientation changes is partitioned to obtain the grain orientation transition area data block.

[0014] Based on the grain orientation transition region data block, local lattice image features in adjacent segments are extracted, and the spatial correspondence between the lattice connection position and the crack trajectory in the image features is determined. Feature cross-analysis is performed according to the degree of torsion of the lattice connection region and the intersection state of the crack trajectory, and the segment positions with strong changes in spatial morphology are identified to obtain the set of lattice misconnection interaction regions.

[0015] Based on the location index of the lattice misconnection interaction region set, the point cloud image information of the corresponding region boundary is extracted, and the three-dimensional orientation parameters of the edge points are determined from the point cloud. The orientation parameters of all segments are classified in spatial order, and regions with prominent orientation differences are labeled. The labeled orientation feature partitions are located and mapped with the geometric distribution information of the original region in space to obtain the structure evolution distribution map.

[0016] As a further aspect of the present invention, the process for determining the three-dimensional orientation parameters of the edge points in the point cloud image information is as follows:

[0017] Based on the geometric relationship between the normal and tangent vectors of edge points in point cloud images, and combined with the spatial distribution of edge points in the three-dimensional coordinate system, the local direction vectors of edge points are extracted and the direction is fitted to form a three-dimensional direction parameter set of edge points.

[0018] During the process of classifying the direction parameters in spatial order, similarity judgment is made based on the angle relationship between the three-dimensional direction vectors of adjacent edge points, and edge points whose direction change is within a preset range are grouped into groups with the same direction.

[0019] The labeling process for regions with prominent directional differences is based on the overall fluctuation of the directional parameters of the edge points in each directional group. If the directional change trend shows obvious discontinuity in space, the region to which the directional group belongs is marked as a spatial segment with directional abrupt change characteristics and included in the directional feature partition.

[0020] As a further aspect of the present invention, the process of generating the multi-source stress transmission path diagram is as follows:

[0021] Based on the structural mutation distribution location in the structural evolution distribution map, extract the layer data of the heat diffusion map and stress migration map during the corresponding time period, select the primitive regions in each layer that spatially overlap with the structural mutation location, compare the heat diffusion intensity and stress direction change trajectory of the same primitive region frame by frame according to the timestamp order, and locate the location where the direction switching state occurs in segments to obtain the set of intersection locations of stress change trajectories.

[0022] The stress change trajectory intersection set is called to extract the stress direction and electric field line direction data of the corresponding positions in the thermal diffusion region, and projection matching is performed according to the direction path with the same starting point. Continuous segments with the same starting point and consistent path direction are screened out during the matching process to obtain the thermoelectric direction overlapping segment group.

[0023] Based on the spatial direction data of the continuous path segments in the thermoelectric direction overlapping segment group, the intersection points where the direction turns in each continuous segment are located and processed. The directional connection relationships of all intersection points are merged into a path chain. The topological pattern of the direction line sequence of each merged path is constructed in three-dimensional space to obtain the multi-source stress transmission direction map.

[0024] As a further aspect of the present invention, the process of generating the electrical response transfer path diagram is as follows:

[0025] The method calls the dense turning section in the multi-source stress transmission path map, retrieves the continuous turning paths of the section in the spatial coordinate system, counts the frequency of direction change of each path, and treats the path area with a frequency higher than the average value of the turning distribution as the direction aggregation area. Based on the spatial location of the direction change points of each path, a direction dataset is constructed to obtain the cluster of turning overlapping paths.

[0026] Based on the aforementioned cluster of overlapping turning paths, the trajectory information that spatially overlaps with the path region in the current mutation trajectory map and the extended channel map of the discharge image is obtained. The trajectory direction sequence and extension vector in each layer are extracted. Cross-screening is performed by the continuity of spatial distance between trajectories and the range of change of direction angle. Line segments that repeatedly pass through the same region and have continuous direction are taken as target extraction objects to obtain a set of repeated trajectory intersection line segments.

[0027] Based on the set of repeated trajectory intersection segments, the angular distribution characteristics of the segments in the spatial coordinate system are grouped into directional vectors. The overall path direction is sorted by the aggregation center coordinates of each group of vectors. The spatial path is labeled according to the density of the vectors. The label information and the spatial coordinate mapping result are converted into directional encoding to obtain the electrical response transfer path diagram.

[0028] As a further aspect of the present invention, the formation process of the aging behavior mapping group is specifically as follows:

[0029] Based on the intersection area of ​​the trajectory distribution in the electrical response transfer path diagram, the corresponding layer of this area in the material peeling record diagram and crack connectivity diagram is retrieved. The continuity state of the pixel blocks with color value jumps and the crack edge contour in the image pixel matrix is ​​extracted. The edge gray value difference and crack curvature change in each layer are processed for segment recognition to obtain the abnormal contour positioning segment set.

[0030] Call the abnormal contour positioning segment set, perform contour line alignment comparison operation on the spatial shape of the edge contour of each segment in the layer and the boundary continuity direction, and group all boundary lines based on contour continuity, and cluster the primitives with high frequency of repeated edge structure states into a group to obtain the boundary cooperation region set.

[0031] Based on the position coordinates of the spatial boundaries of each primitive in the boundary coordination region set, the similarity of the boundary point distribution characteristics between adjacent regions is compared. Pieces with similar structural boundary distribution patterns in the layer are grouped into the same type of map group, and the map group is mapped to the original structural layer to form a spatial aggregation region framework, thereby obtaining the aging behavior mapping group.

[0032] As a further aspect of the present invention, the process of segment identification processing for the difference in edge gray values ​​and the abrupt change in crack curvature in the layer is specifically as follows:

[0033] In the image pixel matrix, a sliding window method is used to perform gradient direction statistics on the neighborhood of each pixel point, extract the direction consistency information of gray value changes in continuous pixel regions, and based on the distribution law of local extreme points of gray value changes on the crack contour direction, combined with the continuity index of curvature change trend, the region where the curvature of the image edge changes abruptly is defined as the abnormal point set.

[0034] In the process of performing contour line alignment comparison on the spatial shape and continuous direction of the boundary of each segment in the layer, the difference in tangent direction and the rate of curvature change of the edge contour line are used as the matching benchmark for alignment comparison. By calculating the consistency coefficient of directional change between adjacent contour points, the contour alignment mapping relationship is constructed.

[0035] The process of clustering primitives with high recurrence frequency of edge structure states into a group is based on the distribution density of the frequency of primitive occurrence in the contour feature space. The density clustering algorithm is used to identify the repetition patterns of primitives in the boundary coordination features, and the clustering results are assigned to the boundary coordination region set.

[0036] As a further aspect of the present invention, the method further includes:

[0037] Based on the spatial block positions in the aging behavior mapping group, check the path superposition in its running trajectory map, determine whether there is overlap between peeling and electrical traces, screen areas with significant trajectory continuity and synchronous damage, classify synchronous aging areas, and output the conclusions on the running status trend and aging mechanism stage performance.

[0038] The conclusions regarding the phased manifestations of the operational status trend and aging mechanism include the distribution of overlapping path areas, the division of material damage and trajectory synchronization areas, and the identification of areas where electrical traces and peeling work together.

[0039] As a further aspect of the present invention, the output process of the conclusions on the operating status trend and aging mechanism stage is as follows:

[0040] Based on the spatial tile positions in the aging behavior mapping group, the trajectory path coordinate information of the corresponding area in the running trajectory record map is extracted, and the path range is located according to the tile spatial boundary. The number of overlaps of all trajectory paths within the tile range is retrieved and processed, and the areas where the path segments appear frequently in time sequence are selected to obtain the trajectory repetition coverage distribution block.

[0041] The trajectory repeat coverage distribution block is invoked to check the spatial coverage of the material peeling record map and the electrical trace fixed area layer in the area. The positioning coordinates of the edge contours in the two types of layers are superimposed with the trajectory path. The material peeling range in the continuous area of ​​the path is synchronously aligned with the light intensity change segment in the electrical trace image to obtain the trajectory material coupling area set.

[0042] Based on the set of trajectory material coupling regions, time series data frames of trajectory change trend and material layer change state are extracted in each region. Frame segments with dense trend transfer points and synchronous layer change amplitude are grouped into the same group. The trajectory behavior characteristics and structural evolution process in all regions of the same group are divided into stages to obtain the stage performance conclusions of the running state trend and aging mechanism.

[0043] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0044] In this invention, by extracting the grain arrangement direction and the abrupt changes in the morphology of the connecting bands, a spatial aggregation layer of structural morphology is established. Combining the coincidence relationship between the thermal diffusion and stress migration paths, a stress input and conduction path map is constructed. By identifying the spatial overlap segments of the current trajectory and the discharge channel, the repeated trajectory areas and the dense distribution of turning directions are marked. Furthermore, material peeling and crack abrupt change boundaries are superimposed to divide the structurally continuous aging block units. By matching and classifying the spatial trajectory and damage features, the synchronous damage areas and staged aging behaviors in the operating path are extracted, enhancing the spatial correlation identification ability between microstructural changes and operating behaviors, and supporting the staged assessment of the operating status trend of electrical equipment and the attribution analysis of the response path. Attached Figure Description

[0045] Figure 1 This is a flowchart of the method of the present invention;

[0046] Figure 2 This is a flowchart illustrating the process of obtaining the structural evolution distribution map of the present invention.

[0047] Figure 3 This is a flowchart illustrating the process of obtaining the multi-source stress transmission path diagram of the present invention.

[0048] Figure 4 This is a flowchart illustrating the process of obtaining the electrical response transfer path diagram of the present invention.

[0049] Figure 5 This is a flowchart illustrating the process of obtaining the aging behavior mapping group in this invention.

[0050] Figure 6 This is a flowchart illustrating the process of obtaining conclusions regarding the operational status trend and the phased performance of the aging mechanism in this invention. Detailed Implementation

[0051] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0052] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0053] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0054] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0055] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0056] Please see Figure 1 This invention provides a technical solution: a method for evaluating the operating status and aging mechanism characteristics of GIS basin insulators, comprising the following steps:

[0057] S1: Retrieve images and material response information of the ceramic body area and conductive contact area at the edge of the GIS basin insulator, analyze the grain arrangement direction and the continuity of the morphology of the connecting band, screen structural segments with lattice interruption and crack intersection in adjacent areas, compare the abrupt trend of structural morphology at the boundary of each segment, superimpose different orientation markers for spatial aggregation and classification, and generate a structural evolution distribution map.

[0058] S2: Based on the distribution location of structural abrupt changes in the structural evolution distribution map, extract the time sequence layers of the heat diffusion map and stress migration map during the operation period, analyze the intersection of the structural abrupt change area and the stress change trajectory, screen the continuous overlapping segments of stress guidance and electric field line direction in the heat diffusion area, record the intersection points of the continuous direction as the effective stress input path, and form a multi-source stress transmission direction map.

[0059] S3: Call the dense turning section in the multi-source stress transmission path diagram, compare the current change trajectory diagram and the discharge image extension channel diagram in this position, determine the degree of overlap of multiple breakdown directions in the spatial path, filter the trajectory intersection segments that appear repeatedly in space, mark the path direction distribution characteristics of the spatial trajectory concentration, and generate the electrical response transfer path diagram.

[0060] S4: Based on the intersection area of ​​trajectory distribution in the electrical response transfer path diagram, retrieve the material peeling record diagram and crack connectivity diagram in the same area, determine the range of surface color difference superposition and crack edge abrupt change in each layer, filter the concentrated area of ​​physical damage traces and unify the layer boundary, divide the boundary block set with consistent structure, and form an aging behavior mapping group.

[0061] S5: Combine the spatial block positions in the aging behavior mapping group to check the overlapping of repeated paths in the operation trajectory record map, determine whether there is a phenomenon of continuous layering and peeling and electrical tracking fixed area overlap, screen areas with significant continuity of operation trajectory and synchronous material damage within the overlapping coverage area, classify the areas where operation behavior and structural aging are synchronous, and output the conclusions of operation status trend and aging mechanism stage performance.

[0062] The structural evolution distribution map includes partitions of grain arrangement direction changes, identification of boundary morphology abrupt change types, and location of lattice interruption and crack intersection points. The multi-source stress transmission path map includes the distribution of thermal-stress coupling paths, effective stress transmission nodes, and sections where stress and electric field directions are consistent. The electrical response transfer path map includes the distribution of overlapping current trajectories, discharge channel intersection characteristics, and dense spatial path sections. The aging behavior mapping group includes concentrated surface delamination patches, crack abrupt change boundary combinations, and color difference superimposed region sets. The conclusions on the operational status trend and the stage performance of the aging mechanism include the distribution of path repetition superimposed areas, the division of material failure and trajectory synchronous areas, and the identification of areas where electrical traces and delamination cooperate.

[0063] Please see Figure 2 The steps to obtain S1 are as follows:

[0064] S101: Retrieve images and material response information of the ceramic body area and conductive contact area at the edge of the GIS basin insulator, perform vector description of the grain orientation information in the image, and extract various response data related to the grain orientation from the material response information. Based on the continuity of the grain orientation vector change in the regional image, perform directional analysis on the orientation state of the connecting strip edge. At the same time, combined with the contact distribution state of the adjacent area edges, perform partitioning processing on the boundary segments where the orientation changes, and obtain the grain orientation transition area data block.

[0065] High-resolution images of the edge ceramic body area and conductive contact area are acquired using industrial vision equipment. Coordinate mapping is performed based on image resolution and actual size to set the boundary position of the recognition area. Edge contours are identified using grayscale gradients, and then the grain structure image of the corresponding area is extracted. Grain boundaries are extracted in areas of grayscale value variation. Based on the direction of grain grayscale texture change, direction vector fitting is performed on each grain region. The principal direction vector of each grain texture direction is obtained through principal gradient direction calculation. By traversing all grain regions in the image and recording the vector direction value and coordinate information of each region, the overall vectorized expression of grain direction in the image is achieved. For example, for a certain ceramic body image region, its principal grain direction is 45°, and the edge grains show continuous changes within a 5-pixel range. The direction change is recorded as 45° to 60°, and this range is marked as the grain transition region. Simultaneously, various performance parameters in the material response information, such as local stress σ, strain ε, and microcrack density ρ, are matched with the region to which the grain direction belongs. By comparing the grain direction with the corresponding material response, the overall direction of the grain structure is obtained. Extract the stress gradient distribution characteristics in the directional change section. For example, in a certain region, the stress changes from 32 MPa to 54 MPa when connecting 45° grains to 60° grains. Further calculate the degree of directional continuity based on the angle between grain vectors of adjacent regions. Regions with an angle change of no more than 10° are defined as continuous regions, while those with an angle greater than 20° are abrupt change regions. Then, perform fitting analysis on the extension path of the connecting zone edge along the grain direction vector, extract its boundary line morphology, and calculate the directional change rate Δθ / Δx. If the change rate is greater than 2° / Δx, the directional change is considered complete. If the value is μm, it is considered to indicate a significant abrupt change in orientation. Further, by combining the overlap rate of adjacent grain boundaries in the conductive contact area image, the boundary segments with abrupt change characteristics are divided into regions. The boundary length, overlap point density, and contact angle change value are extracted to divide the grain transition blocks corresponding to the abrupt change segments. For example, for a transition zone with a length of 35μm, an overlap point density of 8 overlap points within 10μm, and a contact angle that changes from 30° to 75°, this segment is defined as a region of drastic change in grain orientation and is output as a data block for subsequent analysis.

[0066] S102: Based on the grain orientation transition region data block, extract the local lattice image features in adjacent segments, and determine the spatial correspondence between the lattice connection position and the crack trajectory in the image features. Perform feature cross-analysis according to the degree of torsion of the lattice connection region and the intersection state of the crack trajectory, and identify the segment positions with strong changes in spatial morphology to obtain the set of lattice misconnection interaction regions.

[0067] Local regions with a width of 10μm and a length of 20μm were cropped on both sides of the transition zone using image coordinate mapping. Features of the lattice structure within these regions were extracted, and the local image grayscale distribution was transformed into frequency domain texture features. The lattice image features were characterized by three parameters extracted from the two-dimensional Fourier amplitude distribution: the dominant frequency direction, the mean intensity, and the energy concentration. In the example, if the dominant frequency direction is 60°, the mean intensity is 0.72, and the energy concentration is above 85%, it is recorded as having clearly ordered lattice direction features. Further analysis was performed on the lattice intersections within the extracted regions. The lattice connection positions are calculated based on the linear alignment and angular continuity between lattice intersections. If the distance between connection points is less than 1.5 μm and the angle change of the connection direction does not exceed 10°, then a continuous lattice connection is considered to exist. Next, grayscale fracture boundaries are extracted from the crack trajectories appearing in the region. Based on the pixel gradient distribution of the crack boundary lines, the main crack direction line is constructed. If there is an intersection point between the crack direction and the lattice connection direction with an angle of less than 15° and the grayscale abrupt gradient at the intersection point is greater than 20%, it is recorded as a connecting crack intersection point. The coordinates of this intersection point are then recorded and compared with the lattice connection direction. The Euclidean distance between the coordinates of the lattice connection nodes is used to evaluate their spatial correspondence. If the distance between the intersection point and the connection node is less than 2 μm, the crack path is considered to be strongly correlated with the lattice connection. Then, the degree of torsion in the lattice connection region is calculated for each block. The standard deviation of the angle between adjacent connection segments in each lattice connection path is set as the torsion index. If the standard deviation is greater than 12°, it is judged as a region with strong torsion. At the same time, this value is compared with the deflection angle of the crack path. If the difference between the two is within ±10°, it indicates that there is structural directional coupling. The spatial relationship is recorded. The intersection locations were identified, and then the degree of intersection of all fragment regions was comprehensively scored. The scoring criteria were the weighted average of the standard deviation of lattice torsion, crack deflection angle, spatial distance, and intersection density, with each weight set to 0.3, 0.3, 0.2, and 0.2. The weighting was based on the dominant role of the deflection angle in the crack propagation path during the crack deflection sensitivity experiment. Finally, fragment regions with a score greater than 0.6 were selected as fragments with strong spatial morphological changes. The image positions and numbers of these high-scoring regions were then extracted to form a set of lattice misalignment interaction regions.

[0068] S103: Based on the location index of the lattice misconnection interaction region set, extract the point cloud image information of the corresponding region boundary, determine the three-dimensional orientation parameters of the edge points from the point cloud, classify the orientation parameters of all segments in spatial order, and label the regions with prominent orientation differences. Map the labeled orientation feature partitions with the geometric distribution information of the original region in space to obtain the structure evolution distribution map.

[0069] The system retrieves the 3D coordinate range corresponding to each region in the location index, extracts the point cloud image information of the boundary region within this range, and calculates the relative positional relationship of the boundary region points in the 3D coordinate system to form a local neighborhood centered on each edge point. The local neighborhood is defined as a spherical region with a radius of 1.0 μm centered at the edge point. Within this neighborhood, the normal vector direction of the points is calculated. The normal vector is obtained by least-squares plane fitting of the point set within the neighborhood, and the plane normal vector is used as the edge point normal vector. Then, the tangent direction formed between each edge point and its adjacent points is calculated as the tangent vector direction. The direction vector of the edge point in 3D space is extracted and expressed as a unit direction vector group. This direction vector is then smoothed to suppress the influence of local noise. After obtaining the 3D direction parameters of all segment regions, the direction parameters are sorted according to the order of the region spatial coordinates. Directional similarity is judged based on the 3D direction angle between edge points. The judgment criterion is that an angle less than 12° is considered consistent in direction and grouped into the same direction group. If the directional angles of 10 consecutive edge points in a region are all less than 12°, the region is marked as a directionally stable region. If the directional angle between any edge point and the previous direction is greater than 25°, and there are directional abrupt changes of more than 20° in the subsequent 3 points, it is marked as a directionally abrupt change region. Then, the directional trend is fitted to the 3D directional parameters of the edge points within each directional group, and the standard deviation of its directional vector change is judged. If the standard deviation exceeds 15° and the first derivative of the fitted curve shows more than 2 sign jumps, the directional change trend is considered to be discontinuous. The spatial segments corresponding to such groups are marked as having directionally abrupt change characteristics, and labels are attached to their 3D spatial coordinates. The labels use the coordinates of the center point of the region as the positioning markers, indicating the directionally abrupt change level and range of change. Then, all segments with directionally abrupt change labels are mapped to their spatial positions in the original point cloud image, and the directional feature partition is aligned with the geometric position of the original regional structure in space. Through spatial coordinate transformation relationships, a structural evolution distribution map is drawn.

[0070] Please see Figure 3 The steps to obtain S2 are as follows:

[0071] S201: Based on the distribution location of structural mutations in the structural evolution distribution map, extract the layer data of the heat diffusion map and stress migration map during the corresponding time period, select the primitive areas in each layer that spatially overlap with the location of structural mutations, compare the heat diffusion intensity and stress direction change trajectory of the same primitive area frame by frame according to the timestamp order, and locate the location where the direction switching state occurs in segments to obtain the set of intersection positions of stress change trajectories.

[0072] For each abrupt change location in the distribution map, its spatial three-dimensional coordinate index is extracted. Based on the correlation between the coordinates of each abrupt change point and the time axis in the operational data, a frame-by-frame extraction operation is performed on the layer data of the heat diffusion map and stress migration map within the same time period. For each frame, primitive identification is performed on a cubic region with a side length of 5μm centered on the coordinates of the abrupt change point. The heat diffusion intensity value and stress direction vector value of all primitives in this region are extracted. The heat diffusion intensity is recorded after normalizing the infrared radiation intensity value in the layer. The normalization range is set to [0,1]. In the actual example, if the infrared value of a primitive is 175, the maximum value is 255, and the minimum value is 85, the normalization result is (175-85) / (255-85)=0.529. The direction vector is extracted from the arrow direction in the stress migration map. The value of each primitive is set to [0,1]. The direction value is the angle between the arrow direction and the positive X-axis direction, with the direction unit being the angle value. The thermal intensity and stress direction values ​​at each time frame are recorded. The stress direction changes in the same primitive area are compared frame by frame in the order of timestamps. The difference between the stress direction values ​​of two adjacent frames is calculated. If the direction change value exceeds 30°, it is recorded as a direction switching point. At the location of the direction switching point, a continuous change judgment within a 3-frame time window is performed. If the continuous change shows a direction reversal or a deflection rate greater than 15° / frame, the area is determined as a direction change segment. Then, the primitive positions corresponding to the start and end frames of the segment are recorded. Finally, the coordinates of all primitives that have undergone direction changes are organized into a set of stress change trajectory intersection positions, and the maximum thermal diffusion intensity and maximum direction change amplitude at each coordinate point are added to form a complete data output item.

[0073] S202: Call the stress change trajectory intersection set, extract the stress direction and electric field line direction data of the corresponding positions in the thermal diffusion area, and perform projection matching according to the direction path with the same starting point. Filter out continuous segments with the same starting point and consistent path direction during the matching process to obtain the thermoelectric direction overlapping segment group.

[0074] For each intersection point, spatial index matching is performed on the corresponding coordinate position in the thermal diffusion map. The stress direction vector value and electric field line direction vector value of the region coinciding with that coordinate are extracted from the layer. The extraction process maps the vector direction corresponding to the center coordinate of each pixel primitive. Two vectors are set at the same coordinate position as the stress direction vector Ds and the electric field direction vector De, and uniformly represented as unit vectors. For example, if the stress direction is 45°, it is converted to a unit vector (0.707, 0.707). Then, these two direction values ​​are paired according to their respective layer timestamps. Taking each set of starting points as the reference starting point, a direction path is constructed through time windows of 3 frames forward and 3 frames backward. During path construction, the vector trajectory extending from the stress direction and the vector trajectory extending from the electric field line at each point in each frame are connected sequentially to form two direction path line segment sequences. These two... The path overlap is determined by setting the criteria as follows: the distance between the starting coordinates of the path and the starting coordinates is less than 1 μm and the angle between the starting coordinates and the direction is less than 10°. If the starting coordinates are coincident and the direction is consistent, the direction angle of each frame in the continuous path segment is judged frame by frame. If the change in the direction angle within 3 consecutive frames does not exceed 5°, the path segment is considered to have a consistent direction within that frame interval. The starting frame and ending frame of this continuous segment are recorded, and the direction path within this time interval is marked as a candidate overlapping segment. The above judgment operation is performed on all path segments to obtain a set of several path segments that meet the conditions of coincident starting coordinates and continuous and consistent direction. Then, by a second screening based on the path segment length and the angle between the starting coordinates, segments with a path segment length greater than 3 frames and an angle that is always less than 8° are set as effective thermoelectric direction overlapping segments. Finally, all segments that meet the above conditions are organized into thermoelectric direction overlapping segment groups according to their time index and coordinate position.

[0075] S203: Based on the spatial direction data of the continuous path segments in the thermoelectric direction overlapping segment group, the intersection points where the direction turns in each continuous segment are located and processed, and the direction connection relationship of all intersection points is merged into a path chain. The direction topology of each merged path is constructed according to the direction line sequence of each merged path to obtain the multi-source stress transmission direction map.

[0076] The direction values ​​of the direction vectors in each frame within each segment are extracted in 3D coordinates and arranged sequentially to form a direction sequence. The angle between the direction vectors of adjacent frames in the sequence is calculated frame by frame. If the angle change exceeds 20°, the position is marked as a direction inflection point. The 3D coordinates of the inflection point are then recorded as the intersection point. In each continuous segment, the coordinate information of all intersection points and the corresponding preceding and following direction vectors are statistically analyzed. By comparing the angle and extensibility of the direction vectors at the beginning and end of these intersection points, the starting direction is connected to the direction tail of the next intersection point. The criteria for connection are an angle less than 15° and an intermediate path length less than 10μm. Direction segments between two points that meet this condition are classified as the same chain path. Through multiple rounds of iterative judgment, all intersection points are merged according to their direction connection relationship to form a path chain combination. The chain combination structure records the starting coordinates, the number of path segments, and the coordinates of each segment. The system records the spatial orientation of all path chains, including direction values ​​and segment connection order. It projects the direction vector of each direction segment from its 3D direction value into a 3D vector line segment and connects these line segments sequentially according to the path order. A complete path topology is constructed in the 3D coordinate system, where each node is an intersection point and each line segment is a direction segment. For example, if a chain path starts at (5,10,3) and passes through four direction connections (0.5,0.5,0.7), (0.4,0.6,0.7), (0.3,0.6,0.8), and (0.2,0.7,0.7), it forms a polyline structure in 3D space. The average turning angle between the polylines is calculated as a path complexity index. Finally, all chain path sets are numbered, and the path number and orientation information are labeled in the topology diagram, outputting a complete multi-source stress transmission orientation diagram.

[0077] Please see Figure 4 The steps to obtain S3 are as follows:

[0078] S301: Call the dense turning section in the multi-source stress transmission path map, retrieve the continuous turning paths of the section in the spatial coordinate system, count the frequency of direction change of each path, treat the path area with frequency higher than the turning distribution mean as the direction aggregation area, construct the direction dataset based on the spatial location of the direction change points of each path, and obtain the cluster of turning overlapping paths.

[0079] The coordinate sequences of three-dimensional orientation abrupt changes are extracted from all path segments in the image. Spatial blocks with more than 10 orientation abrupt change points within a unit cubic region (10μm side length) are selected and marked as densely turning segments. Then, within each marked segment, the path set is traversed according to the path identifier code. The orientation vector sequences in each path are compared segment by segment, recording whether the angle between each orientation vector and its adjacent segment exceeds a set angle threshold of 20°. If it exceeds, it is determined as an orientation change event. The orientation change events within each path are accumulated and counted. For example, if there are 8 angle changes exceeding the threshold in path P1, the orientation change frequency is 8. After completing the orientation change frequency statistics for all paths, the orientation change frequencies of all paths within the current segment are summarized, and their average frequency is calculated. Paths with a frequency higher than the mean are considered high-frequency direction change paths. The spatial regions covered by these paths are extracted as direction aggregation regions. Then, the coordinates of the points where direction changes occur in each high-frequency path are sampled, their positions in three-dimensional space are extracted and recorded as a set of direction change points, and all points are aggregated to form a direction dataset. The number of times the points with repeated coordinates appear in the dataset is counted and their spatial clustering degree is recorded. If a point appears repeatedly in more than 5 paths and the distance between its adjacent direction change points is less than 3μm, it is determined to be a turning overlap point. Then, the regions where repeated turning points appear between each path are clustered. The clustering is based on the distance being less than 5μm and the number of points being no less than 4. Finally, the high-density clustering of turning overlap path clusters formed by the high-density aggregation of each direction change point is output.

[0080] S302: Based on the cluster of overlapping turning paths, obtain the trajectory information that spatially overlaps with the path region in the trajectory map of current change and the extended channel map of discharge image. Extract the trajectory direction sequence and extension vector in each layer. Perform cross-screening by the continuity of spatial distance between trajectories and the range of change of direction angle. Take the line segments that repeatedly pass through the same area and have continuous direction as the target extraction objects to obtain the set of repeated trajectory cross line segments.

[0081] The spatial boundary range of each cluster in the path cluster set is read, and each cluster is defined as a 3D bounding box in the form of a cubic region with a side length of 10μm. Then, the layer data of the current change trajectory map and the extended channel map of the discharge image are called, and the 3D coordinate point sequence of all trajectory line segments is extracted in the order of layer frames to generate a set of directional paths. Each path line segment records its starting point coordinates, ending point coordinates, and direction vector information. The direction vector is defined as the unit vector value of the first and last points of the path in the 3D coordinate system. Then, the spatial overlap judgment is performed between the spatial bounding boxes of the above trajectory set and each path cluster set. The judgment criterion is that the midpoint coordinates of the trajectory line segment fall within the spatial range of the cluster. If so, the trajectory and the path cluster are considered to have spatial overlap. Then, the directional sequence and direction vector group are extracted from all overlapping trajectories. The spatial overlap segment judgment is performed between each pair of trajectory line segments. The judgment condition is that the distance between the center points of the two trajectory segments is less than 2. The distance between the two points is μm and the angle between the direction vectors is less than 15°. The angle between the direction vectors is calculated by taking the dot product of the unit vectors and then the inverse cosine. For example, if the direction of trajectory segment A is (0.6, 0.8, 0) and the direction of trajectory segment B is (0.61, 0.79, 0.01), the angle between them is 3.5°, which meets the requirement of directional continuity. If the distance between the center points of the two points is 1.6μm, it also meets the requirement of spatial proximity. Then, this segment is recorded as a matching segment. All pairwise matching results of trajectory segments are traversed to filter out the set of line segments that repeatedly pass through the same cluster space range and are directionally continuous. Line segments that appear more than twice at the same position in the set are marked. If the same line segment has a consistent extension trend in both the current trajectory diagram and the discharge diagram, and appears repeatedly in at least three frames, it is recorded as a repeated trajectory intersection line segment. Finally, all matching results that meet the conditions are summarized to form a set of repeated trajectory intersection line segments.

[0082] S303: Based on the set of repeated trajectory intersection segments, perform directional vector grouping processing on their angular distribution characteristics in the spatial directional coordinate system, and sort the overall path direction by combining the aggregation center coordinates of each group of vectors. Assign labels to the spatial path division according to the vector density, and perform directional encoding conversion on the label information and spatial coordinate mapping results to obtain the electrical response transfer path diagram.

[0083] The spatial direction vectors of each line segment are uniformly transformed into unit vectors, and their three-dimensional coordinates are recorded. Then, direction grouping is performed based on the angle between each unit vector. Vectors with an angle less than 10° are grouped into the same group. During direction grouping, the angle θ between the vector and the positive Z-axis is used as a sorting reference. Each vector is compared with the angle of the center vector of an existing group. If the angle is less than 10°, it is added to the current group; otherwise, a new group is created, and the vector becomes the center of the new group. After grouping all vectors, the aggregation center coordinates of the vectors within each group are calculated. The aggregation center is obtained by weighted averaging of the starting coordinates of the line segments within each group. The weights are set according to the length of the line segments; the longer the length, the greater the weight. For example, if line segment L1 has a length of 12μm and its starting point is (4,6,8), and L2 has a length of 8μm and its starting point is (6,5,7), then the aggregation center is calculated as [(12×(4,6,8)+8...]. [×(6,5,7))÷(12+8)]=(4.8,5.6,7.6), then sort all direction vector groups according to the coordinate values ​​of the aggregation center in the Z-axis direction from smallest to largest to obtain the sequence order of the path direction. Assign a unique label code to each group of paths after sorting. The label is in the format of letter + sequence number, such as G1, G2, G3, etc. At the same time, perform a density judgment based on the density of each group of vectors. The density judgment method is the number of paths in a unit cube (1μm³). If the number of paths exceeds 15, it is defined as a high-density segment. Add the suffix mark "H" to the label of the high-density group, such as G2H. After completing the labeling, bind the spatial coordinate point of each path segment to its corresponding label to form a label-coordinate mapping table. Perform a traversal operation on the mapping table to redraw all line segments as direction-coded sequence paths in three-dimensional space. Distinguish all paths by direction label color and connect them in space. Finally, integrate them to form a complete electrical response transfer path diagram.

[0084] Please see Figure 5 The steps to obtain S4 are as follows:

[0085] S401: Based on the intersection area of ​​trajectory distribution in the electrical response transfer path diagram, retrieve the corresponding layer of this area in the material peeling record diagram and crack connectivity diagram, extract the continuity state of the pixel blocks with color value jumps in the image pixel matrix and the crack edge contour, perform segment recognition processing on the edge gray value difference and crack curvature abrupt change in each layer, and obtain the abnormal contour positioning segment set.

[0086] The bounding area of ​​the intersection region in the three-dimensional spatial coordinate system is extracted. Each intersection region is defined by a spatial cube with a side length of 5μm. The coordinate range of this cube is matched with the coordinate index in the material peeling record map and the crack connectivity map. The pixel matrix data of the corresponding layer of the matched region is extracted. The RGB color value of each pixel is differentially analyzed with the color values ​​of its surrounding 3×3 neighborhood. The difference value is set as the sum of the absolute differences. If the difference value exceeds 40, it is recorded as a pixel with a color value jump. The color value threshold of 40 is set by taking the median between the average difference fluctuation of the non-peeled area in the layer (approximately 18) and the difference fluctuation of the peeling edge (approximately 55). Example pixel (120, 118, 11) is used. 5) Compared with neighboring pixels (160, 140, 120), the difference is |120-160|+|118-140|+|115-120|=77. If it exceeds the threshold, it is recorded as a jump point. The pixel blocks formed by the jump points are aggregated according to their adjacency to form color value abnormal blocks. Then, crack edge extraction is performed on the same spatial region in the crack connectivity graph. By traversing along the crack edge point by point and recording its gray value, the gray value difference between pixel i and pixel i+1 is calculated. If the difference value continues to exceed 20 and the direction gradient direction continues to be in the same direction, it is regarded as a continuous segment of crack edge. If the gray value difference changes from positive to negative at a certain pixel point and the absolute value of the rate of change exceeds 25, it is recorded as a crack curvature abrupt change. Points are identified and their coordinates are stored in the crack abrupt change point set. Based on this, segment identification is performed. The continuous crack edge is segmented according to the interval of curvature abrupt change points. The edge segment between each two abrupt change points is considered a candidate segment. The curvature difference at the point of maximum curvature change in each segment is calculated. If the amplitude exceeds 30°, the current segment is marked as an abnormal contour segment. Simultaneously, a sliding window operation is performed on the edge regions of each layer. The window size is set to 5×5 to detect the consistency of grayscale gradient direction within the window. The gradient direction angles of all pixels within the window are counted. If the difference between the maximum and minimum angles is less than 12°, the direction is consistent; if it exceeds 25°, the window is considered to have a direction abrupt change. In the example, if there are 10 pixels within the window... The gradient directions of the elements are 42°, 45°, 47°, 44°, etc., and the difference in direction is within 5°, which is judged as a consistent region. If the value is 70°, the difference becomes 28°, which is judged as a sudden change region. The sudden change region and the color value jump region are cross-matched, and their overlapping positions are taken as the potential anomaly point set. The contour line alignment comparison operation is performed on the potential anomaly points. During the comparison, the tangent direction of the crack edge is used as the reference, and the difference in tangent direction between two adjacent points, Δθ, is recorded as the direction change value. If Δθ is less than 10°, the direction is considered consistent. If Δθ exceeds 20° and there are more than three consecutive points, the position is a direction discontinuity position, and the direction change consistency coefficient is recorded as 1−(Δθ / 90). If the coefficient is greater than 0.Eight alignments were successfully completed. All successfully aligned contour points were linked together to form a mapping chain. The frequency of recurring edge feature points across layers was recorded, and this frequency was used as a clustering criterion, with a frequency threshold set to at least 5 occurrences. Based on the density distribution of frequency in the 3D boundary feature space, clustering was performed using neighborhoods less than 2 μm as the basis. High-density primitives were grouped into boundary co-location regions. For example, if a region contains 26 contour points, and 14 of these points are less than 2 μm apart and appear more than 6 times in multiple layers, then this region is classified as a co-location contour clustering region. Finally, all clustered regions were output as a set of abnormal contour localization segments.

[0087] S402: Call the abnormal contour location segment set, perform contour line alignment comparison operation on the spatial shape of the edge contour of each segment in the layer and the boundary continuity direction, and group all boundary lines based on contour continuity, and cluster the primitives with high frequency of repeated edge structure states into a group to obtain the boundary cooperation region set.

[0088] The spatial coordinate range of each segment in the segment set is extracted. The three-dimensional coordinates of the segment are limited to a local area with a side length of 5μm. The pixel matrix of this coordinate area is retrieved from the corresponding layer of the material peeling record map and the crack connectivity map as the source data for the comparison contour. The edge contour lines within the area are serialized and arranged sequentially according to the pixel number. The coordinates and local tangent direction of each contour point are recorded. The tangent direction is obtained by comparing the coordinate difference between the current point and the adjacent points before and after it. For example... Point Pi has coordinates of (21, 35) and Pi+1 has coordinates of (23, 38). Therefore, the direction angle is arctan((38−35) / (23−21))≈56°. The two contour lines are then aligned. During the comparison, the difference in tangent direction Δθ between corresponding points is used as the alignment criterion. When Δθ is less than 12°, it is recorded as aligned. If more than 8 consecutive points satisfy the direction alignment requirement, this segment is recorded as a continuous contour segment. This continuous segment is marked as a successfully aligned segment in the layer. Next, the alignment is further refined... For successfully aligned segments, region grouping is performed. All contour segments are assigned based on directional consistency and positional proximity. The criteria are set as follows: if the contour direction difference Δθ is less than 15° and the centroid distance between two segments is less than 2μm, they are grouped into the same region group. During the grouping process, the recurrence frequency of each graphic element on the contour line is detected one by one. The frequency is obtained by summing the number of times the same spatial coordinate point appears in each frame layer. In the example, if a graphic element point A is detected in the contour region in 5 consecutive frames, its recurrence frequency is 5. Then, all graphic elements are clustered according to frequency. The density threshold of clustering is set to a frequency of no less than 4 times and a maximum distance between points of no more than 3μm. If a set contains 12 graphic element points and the distance between 9 points is less than 3μm, the set is identified as a high-frequency edge structure cluster and is regarded as a region with strong boundary cooperation. This region is included in the boundary cooperation region set. At the same time, the spatial coordinates and corresponding contour numbers of each point in the set are recorded. Finally, a boundary cooperation region set composed of multiple contour cooperation groups is formed.

[0089] S403: Based on the position coordinates of the spatial boundaries of each primitive in the boundary coordination region set, the similarity of the distribution characteristics of the boundary points between adjacent regions is compared. Pieces with similar structural boundary distribution patterns in the layer are grouped into the same type of map group, and the map group is mapped to the original structural layer to form a spatial aggregation region framework to obtain the aging behavior mapping group.

[0090] The 3D coordinates of each primitive in the set are read and stored in buckets according to the region identifier. The boundary point sequence of primitives in each bucket is sorted to ensure that primitives are arranged continuously from low to high coordinates along the same boundary line. Then, a similarity comparison operation is performed on the boundary point distribution characteristics between adjacent regions. During the comparison, the boundary point sets of adjacent regions are used as the comparison objects. The boundary points of the two regions are compared point by point in corresponding order. The spatial distance δp and the difference in boundary orientation angle δθ of each corresponding point are continuously accumulated and judged. If δp is less than 2μm and δθ is less than 12°, the boundary between the two regions is determined to be within the same region. When the distribution characteristics of points are similar, and the number of points continuously satisfying the above conditions exceeds 40% of the total number of boundary points in a region, the two regions are marked as a group of regions with similar structural patterns, and the average boundary direction angle of the two regions is recorded as the directional characteristic index of the group. In the processing example, if 13 out of the 25 boundary points of region A satisfy δp < 2μm and δθ < 12° when compared with region B, then region A and region B are classified into the same group. Then, a spatial merging operation is performed on all groups that are determined to be of the same type. During the merging process, the bounding boxes of all groups are spatially superimposed based on the minimum bounding box of each group of primitives. If two bounding boxes are... If the overlapping volume of the two bounding boxes exceeds 50% of the smaller bounding box volume, they are merged into a larger spatial aggregate block. The coordinates of the eight vertices of the aggregated bounding box are recorded for subsequent mapping. After completing the spatial aggregation of all groups, coordinate mapping to the original structure layer is performed on each group. The mapping rule is to locate all coordinate points within the bounding box of the aggregate block in the original layer coordinate system and draw the spatial frame boundary of the aggregation region. The boundary is connected to each vertex using a polyline method to form the spatial aggregation region frame on the layer. At the same time, a unique number is assigned to each spatial aggregation frame. The numbers are sorted from low to high according to the Z-axis center coordinate of the spatial aggregation block. For example... For example, if the point density (number of points per 10μm³) in the aggregated region frame exceeds 15, the region is recorded as a dense aging region. If the point density is between 8 and 15, it is recorded as a medium aging region. If it is less than 8, it is recorded as a low-density aging region. The point density threshold is set based on the statistical data in the actual material peeling layer. The average point density of continuous peeling region is about 18, that of slight peeling region is about 9, and that of complete region is about 4. Therefore, 8 and 15 are selected as the interval division basis. Finally, all the aggregated and coded region frames are arranged in numerical order to form aging behavior mapping blocks.

[0091] Please see Figure 6 The steps to obtain S5 are as follows:

[0092] S501: Combine the spatial tile positions in the aging behavior mapping group, extract the trajectory path coordinate information of the corresponding area in the running trajectory record map, locate the path range according to the tile spatial boundary, retrieve the number of overlaps of all trajectory paths within the tile range, filter out the areas where the path segments appear frequently in time sequence, and obtain the trajectory repetition coverage distribution block.

[0093] The spatial boundary coordinates of each aging zone group are located and recorded using a 3D bounding box method. Each tile defines the minimum bounding cube region of its start and end points. For example, the coordinate range of tile AG1 is 20–30 μm for the X-axis, 15–25 μm for the Y-axis, and 5–10 μm for the Z-axis. This range is used as a spatial constraint. All trajectory path data in the trajectory recording map are traversed and retrieved to extract path segments whose start or end coordinates fall within this range. The line segments of the path within the tile are sliced ​​and their spatial position, line segment direction vector, and timestamp are recorded. Then, all the extracted path segments are categorized by tile, and the number of times a path segment in each tile appears repeatedly in all trajectory data is counted. The path overlap frequency is defined as the frequency at which the same line segment is repeatedly traversed at multiple time points. If a path segment P appears 5 times in the AG1 region, its overlap is defined as the frequency at which the same line segment is repeatedly traversed. The number of overlaps is 5. Then, the cumulative statistics of the number of overlaps of all path segments in tile AG1 are performed to calculate the average number of overlaps μ and the standard deviation σ. μ+σ is used as the threshold for judging frequent paths. In the example, if μ is 3.2 and σ is 1.1, the threshold is set to 4.3. Path segments with more than or equal to 5 overlaps are taken as frequent path segments. These path segments are recorded as high-frequency trajectories. Then, the timestamp sequence of high-frequency path segments is sorted to determine whether their time distribution shows repeated trajectory coverage in a continuous period. If the same path segment in the same direction appears more than 3 times in the same area within 10 frames, it is marked as a frequent time series area. Finally, the areas where the path segments that meet the condition of having an overlap number greater than the threshold and have the characteristic of time series repetition are located are clustered into blocks according to the coordinates of the center point of the path segment. The aggregation criteria are that the center point spacing is less than 3μm and the path direction angle is less than 15°. After aggregation, a trajectory repetition coverage distribution block is formed.

[0094] S502: Call the trajectory repeat coverage distribution block, check the spatial coverage of the material stripping record map and the electrical trace fixed area layer in the area, overlay the positioning coordinates of the edge contours in the two types of layers with the trajectory path, and synchronously align the material stripping range in the continuous area of ​​the path with the light intensity change segment in the electrical trace image to obtain the trajectory material coupling area set.

[0095] For each distribution block, a spatial bounding operation is performed on the recorded center coordinates and boundary range, limiting the distribution block to a cube region with a side length of 5μm. Using this spatial region as the search range, the layer data covering the coordinate region is extracted from the material stripping record image. The RGB values ​​and coordinates of all pixels in this region are read, and edge detection processing is performed to obtain the sequence of stripping edge contour coordinate points. The detection method is based on calculating the directional intensity of each point according to the gray-level gradient change. If the gradient value change rate exceeds 20 units of gray-level value, it is marked as an edge point. Similarly, the electrical trace fixing region layer is called, and the light intensity information matrix is ​​extracted within the same spatial coordinate range. The brightness difference of each pixel is calculated and compared with the average value of its surrounding 3×3 neighboring pixels. If the brightness of the current pixel is more than 20% higher than the neighborhood average, it is identified as a light intensity abrupt change point, and its coordinates are added to the abrupt change point set. Subsequently, the stripping edge points and light intensity abrupt change points are superimposed on the trajectory repeating path in the spatial coordinate system. The superposition method is to use the midpoint of the path segment as a reference and extend a buffer of 1μm to both sides. Within the buffer... The number of peeling edge points and light intensity abrupt change points that fall into the data is counted. If the number of points of any one type exceeds 5 and both types of points coexist in the same buffer zone, the current path segment is determined to have material-electrical trace co-occurrence characteristics. The start point, end point, timestamp, and number of overlapping points of the path segment are output as a record item. Then, all path segments with co-occurrence characteristics are classified according to their distribution blocks. The path segment sequence is sorted by time and it is checked whether they form a continuous chain path in the same block. If the distance between the start and end points of the continuous path segments does not exceed 1.5μm and the angle between the direction vectors is not greater than 15°, it is constructed as a continuous path and confirmed as a continuous trajectory region. The synchronous distribution of peeling points and electrical trace points along the path is counted in this region. If there are more than 3 consecutive path segments in the path direction that meet the above co-occurrence conditions and extend in the same direction for not less than 8μm, the corresponding spatial coordinate range is output as the trajectory-material coupling region. The coordinate boundary of this region is selected as a spatial cluster by layer annotation. Finally, the spatial ranges of all path segments that meet the above conditions are integrated to form a trajectory-material coupling region set.

[0096] S503: Based on the trajectory material coupling region set, extract time series data frames of trajectory change trend and material layer change state in each region. Classify frames with dense trend transfer points and synchronous layer change amplitude into the same group. Perform stage division processing on trajectory behavior characteristics and structural evolution process in all regions of the same group to obtain conclusions on the stage performance of operation status trend and aging mechanism.

[0097] The spatial coordinate range of each coupling region is extracted, and all trajectory paths within the region are arranged in time-series path group according to timestamp order. Simultaneously, the grayscale matrix change data of the material layer under the same time series is read. A mapping table is established to correspond the trajectory direction change trend with the material grayscale change trend one-to-one according to time frames. The trajectory direction vector change amplitude Δθ and the material layer grayscale change amplitude Δg in each time frame are recorded. Δθ is obtained by the angle between the direction vectors of adjacent path segments; if the angle is greater than 20°, it is considered a trajectory trend transition point. Δg is the absolute value of the difference between the average grayscale value of the current frame and the average grayscale value of the previous frame. If Δg exceeds the average change in the material region μg (approximately 12 grayscale values) plus the deviation σg (approximately 5 grayscale values) to form a threshold of 17, it is determined to be a sudden material change frame. In the example, if Δg of a certain frame is 23, it is recorded as a sudden change frame. The frame number of the trajectory trend transition point is compared with the material sudden change frame number. If the frame interval between the two does not exceed 2 frames, it is considered a trend synchronization frame. These consecutive segments of synchronization frames are grouped into the same trend group. The trajectory behavior characteristics within all trend groups were then classified. Continuous trajectory direction changes with a length of Δθ less than 15° were defined as stable segments, and continuous segments with Δθ greater than 30° were defined as rapidly changing segments. The time span and spatial path length of each segment were used as the classification criteria. The same processing was applied to material layer changes, with Δg less than 10 considered as slight changes and Δg greater than 20 considered as drastic changes. Next, the combination pattern of trajectory behavior and material changes in each trend group was divided into stages. If a rapidly changing segment of the trajectory exists within a certain time window and the material grayscale change is in the drastic change range, the window is marked as a strongly coupled evolution stage. If the trajectory is a stable segment and the material change is slight, it is marked as a slow evolution stage. If the trajectory shows repeated changes and the material change fluctuates in the medium range (10–20), it is classified as a transitional evolution stage. In this way, a staged evolution sequence is formed for each coupled region. Finally, the stage division results of all regions are organized according to spatial location and temporal order to form a conclusion on the staged performance of the operating status trend and aging mechanism.

[0098] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for evaluating the operating status and aging mechanism characteristics of GIS pot-type insulators, characterized in that, include: Images and material response information of the edge ceramic body and conductive contact area of ​​the GIS basin insulator are retrieved. Grain arrangement and continuity of the connecting bands are analyzed. Structures with lattice interruptions and crack intersections are screened. Boundary abrupt change trends are compared. Spatial clustering is performed by overlaying markers with different orientations to generate a structural evolution distribution map. Specifically, this includes: Images and material response information of the ceramic body area and conductive contact area at the edge of the GIS basin insulator are retrieved. The grain orientation information in the image is vectorized, and various response data related to the grain orientation are extracted from the material response information. Based on the continuity of the grain orientation vector change in the regional image, the orientation state of the connecting strip edge is analyzed. At the same time, combined with the contact distribution state of the adjacent area edge, the boundary segment where the orientation changes is partitioned to obtain the grain orientation transition area data block. Based on the grain orientation transition region data block, local lattice image features in adjacent segments are extracted, and the spatial correspondence between the lattice connection position and the crack trajectory in the image features is determined. Feature cross-analysis is performed according to the degree of torsion of the lattice connection region and the intersection state of the crack trajectory, and the segment positions with strong changes in spatial morphology are identified to obtain the set of lattice misconnection interaction regions. Based on the location index of the lattice misconnection interaction region set, the point cloud image information of the corresponding region boundary is extracted, and the three-dimensional orientation parameters of the edge points are determined from the point cloud. The orientation parameters of all segments are classified in spatial order, and regions with prominent orientation differences are labeled. The labeled orientation feature partitions are located and mapped with the geometric distribution information of the original region in space to obtain the structure evolution distribution map. Based on the distribution location of structural mutations in the structural evolution distribution map, the time series layers of the heat diffusion map and stress migration map are extracted, the intersection relationship between stress trajectory and structural mutation zone is analyzed, the overlapping segments of stress guidance and electric field line direction are screened, the intersection points are recorded, and a multi-source stress transmission direction map is generated. By calling the densely turning sections in the multi-source stress transmission path diagram, comparing the current mutation trajectory diagram with the discharge channel diagram, analyzing the spatial overlap of multiple breakdown directions, filtering repeated intersection trajectories, marking the path direction distribution characteristics, and generating an electrical response transfer path diagram; Based on the intersection area of ​​the electrical response transfer path map, the corresponding material peeling map and crack connectivity map are retrieved to determine the range of surface color difference superposition and crack edge abrupt change, physical damage concentration areas are screened, layer boundaries are unified, and structurally consistent blocks are divided to form aging behavior mapping groups, specifically including: Based on the intersection area of ​​the trajectory distribution in the electrical response transfer path diagram, the corresponding layer of this area in the material peeling record diagram and crack connectivity diagram is retrieved. The continuity state of the pixel blocks with color value jumps and the crack edge contour in the image pixel matrix is ​​extracted. The edge gray value difference and crack curvature change in each layer are processed for segment recognition to obtain the abnormal contour positioning segment set. Call the abnormal contour positioning segment set, perform contour line alignment comparison operation on the spatial shape of the edge contour of each segment in the layer and the boundary continuity direction, and group all boundary lines based on contour continuity, and cluster the primitives with high frequency of repeated edge structure states into a group to obtain the boundary cooperation region set. Based on the position coordinates of the spatial boundaries of each primitive in the boundary coordination region set, the similarity of the boundary point distribution characteristics between adjacent regions is compared. Pieces with similar structural boundary distribution patterns in the layer are grouped into the same type of map group, and the map group is mapped to the original structural layer to form a spatial aggregation region framework to obtain the aging behavior mapping group. The operating status and aging mechanism characteristics of insulators are assessed by analyzing the characteristics of the aging behavior mapping group.

2. The method for evaluating the operating status and aging mechanism characteristics of GIS basin insulators according to claim 1, characterized in that: The structural evolution distribution map includes partitions of grain arrangement direction changes, identification of boundary morphology abrupt change types, and location of lattice interruption and crack intersection points. The multi-source stress transmission path map includes thermal-stress coupling path distribution, effective stress transmission nodes, and stress and electric field direction consistent sections. The electrical response transfer path map includes current trajectory overlap distribution, discharge channel intersection characteristics, and dense spatial path sections. The aging behavior mapping group includes surface peeling concentrated patches, crack abrupt change boundary combinations, and color difference superposition region sets.

3. The method for evaluating the operating status and aging mechanism characteristics of GIS basin insulators according to claim 1, characterized in that, The process of determining the three-dimensional orientation parameters of edge points in the point cloud image information is as follows: Based on the geometric relationship between the normal and tangent vectors of edge points in point cloud images, and combined with the spatial distribution of edge points in the three-dimensional coordinate system, the local direction vectors of edge points are extracted and the direction is fitted to form a three-dimensional direction parameter set of edge points. During the process of classifying the direction parameters in spatial order, similarity judgment is made based on the angle relationship between the three-dimensional direction vectors of adjacent edge points, and edge points whose direction change is within a preset range are grouped into groups with the same direction. The labeling process for regions with prominent directional differences is based on the overall fluctuation of the directional parameters of the edge points in each directional group. If the directional change trend shows obvious discontinuity in space, the region to which the directional group belongs is marked as a spatial segment with directional abrupt change characteristics and included in the directional feature partition.

4. The method for evaluating the operating status and aging mechanism characteristics of GIS basin insulators according to claim 1, characterized in that, Based on the structural abrupt change distribution locations in the structural evolution distribution map, time-series layers of the heat diffusion map and stress migration map are extracted. The intersection relationship between stress trajectories and structural abrupt change regions is analyzed. Overlapping segments of stress guidance and electric field line directions are selected, and intersection points are recorded to generate a multi-source stress conduction direction map, including: Based on the structural mutation distribution location in the structural evolution distribution map, extract the layer data of the heat diffusion map and stress migration map during the corresponding time period, select the primitive regions in each layer that spatially overlap with the structural mutation location, compare the heat diffusion intensity and stress direction change trajectory of the same primitive region frame by frame according to the timestamp order, and locate the location where the direction switching state occurs in segments to obtain the set of intersection locations of stress change trajectories. The stress change trajectory intersection set is called to extract the stress direction and electric field line direction data of the corresponding positions in the thermal diffusion region, and projection matching is performed according to the direction path with the same starting point. Continuous segments with the same starting point and consistent path direction are screened out during the matching process to obtain the thermoelectric direction overlapping segment group. Based on the spatial direction data of the continuous path segments in the thermoelectric direction overlapping segment group, the intersection points where the direction turns in each continuous segment are located and processed. The directional connection relationships of all intersection points are merged into a path chain. The topological pattern of the direction line sequence of each merged path is constructed in three-dimensional space to obtain the multi-source stress transmission direction map.

5. The method for evaluating the operating status and aging mechanism characteristics of GIS basin insulators according to claim 1, characterized in that, By calling upon the densely turning sections in the multi-source stress transmission path diagram, comparing the current mutation trajectory diagram with the discharge channel diagram, analyzing the spatial overlap of multiple breakdown directions, filtering out repeated intersection trajectories, marking path direction distribution characteristics, and generating an electrical response transfer path diagram, including: The method calls the dense turning section in the multi-source stress transmission path map, retrieves the continuous turning paths of the section in the spatial coordinate system, counts the frequency of direction change of each path, and treats the path area with a frequency higher than the average value of the turning distribution as the direction aggregation area. Based on the spatial location of the direction change points of each path, a direction dataset is constructed to obtain the cluster of turning overlapping paths. Based on the aforementioned cluster of overlapping turning paths, the trajectory information that spatially overlaps with the path region in the current mutation trajectory map and the extended channel map of the discharge image is obtained. The trajectory direction sequence and extension vector in each layer are extracted. Cross-screening is performed by the continuity of spatial distance between trajectories and the range of change of direction angle. Line segments that repeatedly pass through the same region and have continuous direction are taken as target extraction objects to obtain a set of repeated trajectory intersection line segments. Based on the set of repeated trajectory intersection segments, the angular distribution characteristics of the segments in the spatial coordinate system are grouped into directional vectors. The overall path direction is sorted by the aggregation center coordinates of each group of vectors. The spatial path is labeled according to the density of the vectors. The label information and the spatial coordinate mapping result are converted into directional encoding to obtain the electrical response transfer path diagram.

6. The method for evaluating the operating status and aging mechanism characteristics of GIS basin insulators according to claim 1, characterized in that, The process of segment identification processing based on the difference in edge gray values ​​and the abrupt change in crack curvature in the layer is as follows: In the image pixel matrix, a sliding window method is used to perform gradient direction statistics on the neighborhood of each pixel point, extract the direction consistency information of gray value changes in continuous pixel regions, and based on the distribution law of local extreme points of gray value changes on the crack contour direction, combined with the continuity index of curvature change trend, the region where the curvature of the image edge changes abruptly is defined as the abnormal point set. In the process of performing contour line alignment comparison on the spatial shape and continuous direction of the boundary of each segment in the layer, the difference in tangent direction and the rate of curvature change of the edge contour line are used as the matching benchmark for alignment comparison. By calculating the consistency coefficient of directional change between adjacent contour points, the contour alignment mapping relationship is constructed. The process of clustering primitives with high recurrence frequency of edge structure states into a group is based on the distribution density of the frequency of primitive occurrence in the contour feature space. The density clustering algorithm is used to identify the repetition patterns of primitives in the boundary coordination features, and the clustering results are assigned to the boundary coordination region set.

7. The method for evaluating the operating status and aging mechanism characteristics of GIS basin insulators according to claim 1, characterized in that, The method further includes: Based on the spatial block positions in the aging behavior mapping group, check the path superposition in its running trajectory map, determine whether there is overlap between peeling and electrical traces, screen areas with significant trajectory continuity and destruction synchronicity, classify synchronous aging areas, and output the running status trend and aging mechanism stage performance conclusions. The conclusions regarding the phased manifestations of the operational status trend and aging mechanism include the distribution of overlapping path areas, the division of material damage and trajectory synchronization areas, and the identification of areas where electrical traces and peeling work together.

8. The method for evaluating the operating status and aging mechanism characteristics of GIS basin insulators according to claim 7, characterized in that, The specific process for obtaining conclusions on the operational status trend and aging mechanism stage performance is as follows: Based on the spatial tile positions in the aging behavior mapping group, the trajectory path coordinate information of the corresponding area in the running trajectory record map is extracted, and the path range is located according to the tile spatial boundary. The number of overlaps of all trajectory paths within the tile range is retrieved and processed, and the areas where the path segments appear frequently in time sequence are selected to obtain the trajectory repetition coverage distribution block. The trajectory repeat coverage distribution block is invoked to check the spatial coverage of the material peeling record map and the electrical trace fixed area layer in the area. The positioning coordinates of the edge contours in the two types of layers are superimposed with the trajectory path. The material peeling range in the continuous area of ​​the path is synchronously aligned with the light intensity change segment in the electrical trace image to obtain the trajectory material coupling area set. Based on the set of trajectory material coupling regions, time series data frames of trajectory change trend and material layer change state are extracted in each region. Frame segments with dense trend transfer points and synchronous layer change amplitude are grouped into the same group. The trajectory behavior characteristics and structural evolution process in all regions of the same group are divided into stages to obtain the stage performance conclusions of the running state trend and aging mechanism.

Images