Intelligent Inspection Method and Device Based on Machine Vision

Abstract

This invention relates to the field of computer technology, and provides an intelligent inspection method and apparatus based on machine vision. The method includes: acquiring images of the visible section of the route to be inspected using a machine vision device to obtain full-domain image data; identifying route anomalies based on the full-domain image data to obtain anomaly location information; analyzing the spatial proximity relationships and physical connection relationships between the various route anomalies based on the anomaly location information to construct anomaly distribution relationships; determining an inspection path based on the first current inspection position of the inspection device and the anomaly distribution relationships to determine a target inspection path; if new anomalies are detected during the inspection process along the target inspection path, updating the target inspection path based on the new anomalies and the anomaly distribution relationships to obtain an updated inspection path, and performing inspection along the updated path. This invention ensures dynamic and efficient inspection operations.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to an intelligent inspection method and device based on machine vision. Background Technology

[0002] In power line inspection, the rationality of the inspection route directly affects inspection efficiency and the comprehensiveness of anomaly detection. The current mainstream inspection method is the fixed-route inspection method. This involves pre-surveying the area to be inspected, planning and storing one or more fixed inspection routes, and then strictly following these routes during the actual inspection to complete the work. The core limitation of the existing fixed-route inspection method is that route planning relies on pre-acquired static information. Once the route is determined, it cannot be flexibly adjusted, completely detached from the dynamic scenarios and actual condition of the power line during the inspection process.

[0003] However, when using the fixed-path inspection method for line inspection, the fixed path cannot dynamically adapt to the actual line conditions, leading to wasted inspection resources, low efficiency in anomaly detection, and the tendency to miss dynamically emerging anomalies. Specifically, the fixed path is based on preliminary static survey information and cannot adapt to the actual distribution of anomalies on the line. This often results in the inspection device repeatedly turning back and forth between scattered anomalies, with excessively long and ineffective travel distances. At the same time, the fixed path cannot cover newly emerging anomalies during the inspection process in a timely manner, requiring a return to check only after completing the entire preset path. This not only delays the handling of new anomalies but also further increases inspection time. Overall, this leads to an imbalance in the allocation of inspection resources, low efficiency in anomaly detection, and an inability to achieve dynamic and efficient inspection operations. Summary of the Invention

[0004] This invention provides an intelligent inspection method and device based on machine vision, which aims to dynamically adapt the inspection path to the actual line status during the inspection process, thereby ensuring dynamic and efficient inspection operations.

[0005] In a first aspect, the present invention provides an intelligent inspection method based on machine vision, comprising:

[0006] The machine vision equipment is used to acquire images of the visible section of the line to be inspected, and full-area image data is obtained. Based on the full-area image data, line anomaly points are identified to obtain anomaly point location information.

[0007] Based on the spatial proximity relationship and physical connection relationship between each line anomaly point obtained from the analysis of the anomaly point location information, the distribution relationship of the anomaly points is constructed.

[0008] Based on the first current inspection position of the inspection device and the distribution relationship of the abnormal points, the inspection path is determined to identify the target inspection path.

[0009] If the inspection device detects new anomalies during the inspection process along the target inspection path, the target inspection path is updated based on the new anomalies and their distribution relationship to obtain the updated inspection path, and the inspection is carried out along the updated inspection path.

[0010] Optionally, the target inspection path is updated based on the newly added anomalies and the distribution relationship of the anomalies to obtain the updated inspection path, including:

[0011] For each newly added anomaly, determine whether it has a physical connection or spatial proximity with any line anomaly in any local anomaly cluster area in the anomaly distribution relationship.

[0012] If it exists, the new anomaly point is added to the corresponding first target local anomaly cluster area, and the boundary points of the anomaly region of the first target local anomaly cluster area are updated based on the new anomaly point to obtain the updated local anomaly cluster area; if it does not exist, the new anomaly point is determined as a new local anomaly candidate area.

[0013] Based on the target inspection path, boundary point coverage detection is performed on the updated local anomaly cluster area and the newly added local anomaly candidate area. Any local anomaly cluster area or newly added local anomaly candidate area whose boundary point is not covered by the target inspection path is determined as an uncovered anomaly area.

[0014] The updated inspection path is obtained by updating the target inspection path based on the second current inspection position of the inspection device on the target inspection path and the uncovered abnormal area.

[0015] Optionally, the target inspection path is updated based on the second current inspection position of the inspection device on the target inspection path and the uncovered abnormal area, resulting in the updated inspection path, including:

[0016] Starting from the second current inspection position, based on the physical connection relationship of the line, the transmission line topology is traversed, and the first local anomaly cluster area or newly added local anomaly candidate area reached through the physical connection relationship of the continuous line in the uncovered anomaly area is determined as the second target local anomaly cluster area.

[0017] Based on the physical connection relationship between the second current inspection position and the boundary point of the abnormal area at the starting end of the second target local abnormal cluster area, the continuous line segment from the second current inspection position to the boundary point of the abnormal area at the starting end is determined as the insertion access path segment;

[0018] If the second target local anomaly cluster area is the updated local anomaly cluster area, then a supplementary traversal path is generated based on the anomaly point sequence path corresponding to the updated local anomaly cluster area; if the second target local anomaly cluster area is a newly added local anomaly candidate area and contains only a single newly added anomaly point, then the line segment where the newly added anomaly point is located is used as the supplementary traversal path; if the second target local anomaly cluster area is a newly added local anomaly candidate area and contains multiple newly added anomaly points, then its anomaly point sequence path is generated according to the line physical connection relationship, and the anomaly point sequence path is used as the supplementary traversal path.

[0019] The updated inspection path is obtained by concatenating the inserted access path segment, the supplementary traversal path, and the unexecuted path after the second current inspection position in the target inspection path.

[0020] Optionally, based on the first current inspection position of the inspection device and the distribution relationship of the abnormal points, an inspection path decision is made to determine the target inspection path, including:

[0021] Based on the distribution relationship of the abnormal points, the local abnormal cluster area that is directly connected to the first current inspection position through the physical connection of the line by the boundary point of the abnormal area is determined as the initially reachable abnormal area.

[0022] Starting from the initial reachable abnormal region, based on the physical connection relationship of the line indicated by the abnormal distribution topology in the abnormal point distribution relationship, the remaining local abnormal clusters directly connected to it are traversed in sequence to obtain the abnormal region adjacency sequence.

[0023] The local abnormal clusters in the abnormal region adjacency sequence are sorted according to their physical arrangement on the transmission line to obtain the abnormal region access order that extends unidirectionally from the initial reachable abnormal region along the physical connection relationship of the line.

[0024] Based on the boundary points of the abnormal regions of two adjacent local abnormal clusters in the abnormal region access sequence, the inspection path is determined to identify the target inspection path.

[0025] Optionally, the inspection path is determined based on the boundary points of two adjacent local anomaly clusters in the anomaly region access sequence, including:

[0026] For two adjacent local abnormal clusters in the abnormal area access sequence, the connection path segment between the regions is determined based on the physical connection relationship between the boundary point of the abnormal area corresponding to the end of the previous local abnormal cluster and the boundary point of the abnormal area corresponding to the beginning of the next local abnormal cluster.

[0027] Based on the physical connection relationship between the first current inspection location and the boundary point of the abnormal area corresponding to the starting end of the initially reachable abnormal area, a continuous line segment from the first current inspection location to the boundary point of the abnormal area corresponding to the starting end of the initially reachable abnormal area is determined to obtain the initial access path segment.

[0028] For each local anomaly cluster in the abnormal region access sequence, bidirectional candidate traversal paths with opposite traversal directions are generated based on its corresponding anomaly point sequence path; the first candidate path traverses from the anomaly region boundary point at the beginning to the anomaly region boundary point at the end, and the second candidate path traverses from the anomaly region boundary point at the end to the anomaly region boundary point at the beginning.

[0029] The inspection path is determined by making inspection path decisions based on the initial access path segment, the inter-regional connection path segments of adjacent local anomaly clusters, and the bidirectional candidate traversal path of each local anomaly cluster.

[0030] Optionally, based on the initial access path segment, the inter-regional connection path segments of adjacent local anomaly clusters, and the bidirectional candidate traversal path of each local anomaly cluster, an inspection path decision is made to determine the target inspection path, including:

[0031] The endpoint of the initial access path segment is matched with the two boundary points of the abnormal region of the initial reachable abnormal region. If the endpoint of the initial access path segment is connected to the boundary point of the abnormal region at the beginning of the initial reachable abnormal region through a physical connection, the candidate traversal path in the traversal direction from the beginning to the end is determined as the target traversal path. If the endpoint of the initial access path segment is connected to the boundary point of the abnormal region at the end of the initial reachable abnormal region, the candidate traversal path in the traversal direction from the end to the beginning is determined as the target traversal path.

[0032] For each local anomaly cluster starting from the second local anomaly cluster in the anomaly region access sequence, the endpoint of the candidate traversal path of the previous local anomaly cluster is compared with the two anomaly region boundary points of the current local anomaly cluster. If the endpoint is connected to the anomaly region boundary point of the starting end of the current local anomaly cluster through a physical connection, the candidate traversal path in the traversal direction from the starting end to the ending end is determined as the target traversal path. If the endpoint is connected to the anomaly region boundary point of the ending end, the candidate traversal path in the traversal direction from the ending end to the starting end is determined as the target traversal path. In this way, the endpoints of the candidate traversal paths are compared with the two anomaly region boundary points to obtain the target traversal paths for the remaining local anomaly clusters.

[0033] The target inspection path is obtained by sequentially connecting the initial access path segment, the target traversal path of each local anomaly cluster area, and the inter-regional connection path segment of adjacent local anomaly cluster areas.

[0034] Optionally, based on the spatial proximity and physical connection relationships between various line anomalies obtained from the anomaly location information analysis, anomaly point distribution relationships are constructed, including:

[0035] For any two different first line anomaly points and second line anomaly points, if there is at least one relationship between the first line anomaly point and the second line anomaly point, either a spatial proximity relationship or a physical connection relationship, then the first line anomaly point and the second line anomaly point are determined as an anomaly point pair.

[0036] Merge pairs of anomalies that share the same line anomaly point to obtain adjacent anomaly point groups; any two line anomalies within an adjacent anomaly point group can be connected to each other through the path in the anomaly point pair.

[0037] For each adjacent anomaly point group, an anomaly point subgraph is constructed with each line anomaly point in the group as a node and the physical connection relationship or spatial proximity relationship between each line anomaly point in the group as an edge.

[0038] The distribution relationship of the abnormal points is constructed based on the route indicated by the physical connection relationship of the route in the connected subgraph of each abnormal point.

[0039] Optionally, the distribution relationship of the abnormal points is constructed based on the route indicated by the physical connection relationship of the lines in the connectivity subgraph of each abnormal point, including:

[0040] Based on the route indicated by the physical connection relationship of the line in the connected subgraph of each anomaly point, adjacent line anomaly points are connected in sequence to obtain the anomaly point sequence path; there is a physical connection relationship between adjacent line anomaly points in each anomaly point sequence path.

[0041] For any target path in the anomaly point sequence path, if the number of line anomalies it contains is greater than or equal to a preset minimum anomaly point threshold, then the line segment covered by the target path is determined as a local anomaly cluster area.

[0042] For each local anomaly cluster, the line anomaly points corresponding to its starting and ending ends are taken as the boundary points of the anomaly region. Combining the connection relationship between each local anomaly cluster in the anomaly point connectivity subgraph, an anomaly distribution topology is constructed. The anomaly distribution topology represents the interconnection relationship between each local anomaly cluster on the line to be inspected.

[0043] The abnormal point distribution relationship is obtained by integrating the local abnormal clusters, abnormal region boundary points and their interconnections contained in the abnormal distribution topology.

[0044] Secondly, the present invention also provides a machine vision-based intelligent inspection device for implementing the machine vision-based intelligent inspection method as described in the first aspect; the device includes:

[0045] The anomaly identification module is used to acquire images of the visible section of the line to be inspected based on machine vision equipment, obtain full-area image data, and identify line anomalies based on the full-area image data to obtain anomaly location information.

[0046] The distribution relationship construction module is used to construct the distribution relationship of abnormal points based on the spatial proximity relationship and physical connection relationship between each abnormal point of the line obtained by analyzing the abnormal point location information.

[0047] The inspection path decision module is used to make inspection path decisions based on the first current inspection position of the inspection device and the distribution relationship of the abnormal points, and to determine the target inspection path.

[0048] The inspection path update module is used to update the target inspection path based on the newly added abnormal points and the distribution relationship of the abnormal points if the inspection device detects that there are new abnormal points during the inspection process along the target inspection path, so as to obtain the updated inspection path, and to carry out inspection along the updated inspection path.

[0049] Thirdly, the present invention also provides an electronic device, comprising: a memory for storing computer software programs; and a processor for reading and executing the computer software programs, thereby realizing the intelligent inspection method based on machine vision as described above.

[0050] Fourthly, the present invention also provides a non-transitory computer-readable storage medium storing a computer software program, which, when executed by a processor, implements the intelligent inspection method based on machine vision as described above.

[0051] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the intelligent inspection method based on machine vision as described above.

[0052] The intelligent inspection method based on machine vision provided in this invention analyzes the spatial proximity and physical connection relationships between anomaly points on various lines based on anomaly point location information, constructing anomaly point distribution relationships and clearly presenting the spatial distribution patterns and correlation characteristics of anomaly points. Therefore, it can plan inspection paths adapted to the actual situation. Based on the anomaly point distribution relationships and the first current inspection position of the inspection device, inspection path decisions are made to determine the target inspection path. This ensures that the target inspection path closely matches the actual distribution and correlation relationships of anomaly points, avoiding repeated back-and-forth trips between scattered anomaly points and excessively long ineffective travel distances. It also reduces ineffective time spent in areas without anomaly points, solving the problems of wasted inspection resources and low anomaly point investigation efficiency. Based on the target inspection path, when a new anomaly point is detected during the inspection process, the target inspection path is updated based on the new anomaly point and its distribution relationship to obtain an updated inspection path. This allows for flexible adjustment of the inspection path according to the dynamic state of the actual line, enabling timely coverage of new anomaly points. It avoids the drawbacks of not being able to respond to new anomaly points in a timely manner and requiring retrospective investigation, solving the problems of easily overlooking dynamically occurring anomalies, delaying processing opportunities, and increasing inspection time. Therefore, the embodiments of the present invention enable the inspection path to dynamically adapt to the actual line status during the inspection process, thereby achieving reasonable allocation of inspection resources, improving inspection efficiency, and dynamically covering constant points, ensuring dynamic and efficient inspection operations. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the intelligent inspection method based on machine vision provided in an embodiment of the present invention.

[0054] Figure 2 This is a schematic diagram of the structure of the intelligent inspection device based on machine vision provided in an embodiment of the present invention;

[0055] Figure 3 An embodiment diagram of the electronic device provided in this invention;

[0056] Figure 4 An embodiment diagram of a computer-readable storage medium provided in accordance with the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0059] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0060] Optionally, see Figure 1 , Figure 1 This is a flowchart illustrating the intelligent inspection method based on machine vision provided by the present invention. In this embodiment, the executing entity of the intelligent inspection method based on machine vision is an inspection device. In one embodiment, the inspection device in this invention may be an inspection robot, an inspection drone, etc. Therefore, the intelligent inspection method based on machine vision includes:

[0061] Step 10: Based on the machine vision equipment, images are acquired from the visible section of the line to be inspected to obtain full-area image data. Based on the full-area image data, abnormal points of the line are identified to obtain the location information of the abnormal points.

[0062] Optionally, the machine vision equipment carried by the inspection device (machine vision equipment refers to equipment with image capture, transmission and preliminary preprocessing functions, such as high-definition industrial cameras, infrared thermal imaging cameras, etc.) systematically scans and collects images of the preset visible section of the line to be inspected. The visible section of the line to be inspected refers to the area within the current inspection range where the inspection device can clearly capture the image of the line through the machine vision equipment, excluding areas that are completely blocked by obstacles or where the lighting conditions cannot meet the imaging requirements.

[0063] Optionally, during the data acquisition process, the inspection device moves smoothly along a preset initial trajectory. The machine vision equipment captures images based on the set acquisition frequency (e.g., twice per second) and imaging parameters (e.g., focal length, exposure time, sensitivity, etc., preset according to factors such as the light intensity of the environment to be inspected, the contrast between the line and the background, etc.) to ensure that the acquired images can completely cover the entire visible area without any omissions.

[0064] Furthermore, the machine vision equipment transmits the acquired single-frame images to the control module of the inspection device in real time. The control module performs preliminary preprocessing on the single-frame images (such as image denoising, size normalization, image stitching and alignment) to obtain full-area image data covering the visible section of the line to be inspected.

[0065] Furthermore, the inspection device invokes a preset anomaly detection algorithm to identify line anomalies based on the acquired full-area image data. The core logic of the anomaly detection algorithm is to extract the actual feature information of the line from the full-area image data (such as the line's contour shape, color distribution, and texture features).

[0066] Among them, contour morphology refers to the edge contour and direction of the line in the image. Color distribution refers to the color depth and uniformity of the line surface; texture features refer to the detailed features of the line surface such as texture and wear marks.

[0067] Furthermore, the inspection device compares the extracted actual feature information with a preset normal line feature template (a feature benchmark dataset trained on a large number of samples based on standard line images without anomalies, containing standard feature parameters such as the contour, color, and texture of normal lines). When the comparison finds a deviation between the actual feature information and the normal line feature template, and the deviation value exceeds a preset allowable threshold (a pre-set critical value for judging whether a feature deviation constitutes an anomaly, determined according to factors such as line type and inspection accuracy requirements), the line location corresponding to the deviation is determined to be an anomaly point.

[0068] Furthermore, the positioning module of the inspection device (a component with location information acquisition function, such as a GPS positioning module, a laser positioning module, etc.) simultaneously acquires the actual spatial location information corresponding to the abnormal point, and obtains the abnormal point location information. The abnormal point location information refers to the information that can accurately identify the specific spatial coordinates of the abnormal point in the line to be inspected, such as three-dimensional coordinate data based on a preset coordinate system.

[0069] In one embodiment, taking the inspection of an overhead power transmission line as an example, the inspection device is an inspection robot, which carries a high-definition industrial camera and an infrared thermal imaging camera as machine vision equipment. The visible section of the line to be inspected is a 100-meter-long section of overhead power transmission line. This section has no tall trees, buildings, or other obstructions, and the lighting conditions are good. The inspection robot moves smoothly along the track beside the power transmission line according to a preset initial trajectory. Both the high-definition industrial camera and the infrared thermal imaging camera are set to a sampling frequency of once per second. The focal length of the high-definition industrial camera is set to 50mm, and the exposure time is set to 1 / 100 second. The sensitivity of the infrared thermal imaging camera is set to 800.

[0070] During the inspection robot's movement, two cameras simultaneously acquire images of the 100-meter visible area, collecting a total of 200 single-frame images. The inspection robot's control module performs preliminary preprocessing on the acquired single-frame images: Gaussian filtering is used to denoise the images, eliminating noise caused by ambient light fluctuations; all single-frame images are uniformly adjusted to a size of 1920×1080 pixels to complete size normalization; adjacent single-frame images are stitched and aligned using the SIFT algorithm (Scale Invariant Feature Transform algorithm, which detects feature points in the image, calculates the descriptors of the feature points, and achieves feature point matching between different images), ultimately obtaining full-domain image data covering the entire 100-meter visible area.

[0071] Subsequently, the control module extracts the outline shape, surface color distribution, and texture features of the transmission line from the global image data. The extracted actual feature information is compared with a preset feature template of a normal overhead transmission line. It is found that in the global image data, the line area 35 meters from the initial position of the inspection robot has a localized protrusion in its outline (the deviation from the smooth outline of a normal line is 0.8 mm, exceeding the preset allowable threshold of 0.5 mm), and the temperature of this area in the infrared thermal imaging image is 15°C higher than that of the normal line area (the temperature deviation exceeds the preset allowable threshold of 10°C). Therefore, this area is determined to be an anomaly point in the line, and the GPS positioning module of the inspection robot simultaneously obtains the three-dimensional coordinates of this anomaly point as (X1, Y1, Z1).

[0072] Step 20: Based on the spatial proximity relationship and physical connection relationship between each abnormal point on the line obtained from the analysis of the abnormal point location information, construct the distribution relationship of abnormal points.

[0073] Optionally, the inspection device calculates the actual spatial straight-line distance (i.e., the length of the line connecting the two anomaly points in a three-dimensional spatial coordinate system) between any two anomaly points based on the anomaly point location information of each line anomaly point. Further, the inspection device compares the calculated actual spatial straight-line distance with a preset proximity threshold (a pre-set critical distance used to determine whether two anomaly points have a spatial proximity relationship, determined based on factors such as the installation density of the line to be inspected and the inspection accuracy requirements). If the actual spatial straight-line distance between two anomaly points is less than or equal to the preset proximity threshold, the two anomaly points are determined to have a spatial proximity relationship; if the actual spatial straight-line distance is greater than the preset proximity threshold, the two anomaly points are determined not to have a spatial proximity relationship.

[0074] Optionally, the inspection device pre-stores physical topology data of the line to be inspected (a dataset recording the route, connection method, joint location, branch information, etc. of each section of the line to be inspected). During analysis, based on the anomaly location information of each line anomaly point, the specific line section where each anomaly point is located is located in the physical topology data. Then, the connection attributes of the line sections where each anomaly point is located are queried in the physical topology data to determine whether the line sections where each anomaly point is located belong to the same continuous line, whether they are directly connected through joints or other components, or whether they are indirectly connected through other line sections. If the line sections where two or more anomaly points are located belong to the same continuous line, or are directly connected through joints, then these anomaly points are determined to have a direct physical connection relationship. If the line sections where two anomaly points are located are indirectly connected through other line sections, then they are determined to have an indirect physical connection relationship. If the line sections where two anomaly points are located have no direct or indirect connection relationship, then they are determined not to have a physical connection relationship.

[0075] Furthermore, the inspection device analyzes the abnormal points of each line based on spatial proximity and physical connection of the lines to determine local abnormal clusters.

[0076] Furthermore, the inspection device determines the boundary points of the abnormal areas in each local abnormal cluster area, as well as the connections formed between different local abnormal cluster areas through physical connections of lines, clarifies their interconnection relationships, and integrates the local abnormal cluster areas, the boundary points of the abnormal areas, and the interconnection relationships of the local abnormal cluster areas to obtain the distribution relationship of abnormal points, as specifically in steps 201 to 204.

[0077] Step 30: Based on the relationship between the first current inspection position of the inspection device and the distribution of abnormal points, make an inspection path decision and determine the target inspection path.

[0078] Optionally, the inspection device makes an inspection path decision based on its first current inspection position (i.e., the actual spatial position of the inspection device) and the distribution relationship of abnormal points. The decision-making process minimizes the inspection path length and reduces the inspection time while ensuring coverage of all local abnormal clusters and boundary points of abnormal areas. At the same time, it prioritizes planning paths that pass through interconnected lines of local abnormal clusters to ensure the continuity and integrity of the inspection, thus obtaining the target inspection path, as described in steps 301 to 304.

[0079] Step 40: If a new abnormal point is detected during the inspection process of the inspection device along the target inspection path, the target inspection path is updated based on the new abnormal point and the distribution relationship of the abnormal point to obtain the updated inspection path, and the inspection is carried out along the updated inspection path.

[0080] Optionally, the inspection device moves along the target inspection path for inspection. During the movement, its machine vision equipment continuously acquires real-time images of the line sections along the path according to the acquisition frequency and imaging parameters set in step 10. The inspection device processes the real-time acquired image data in real time, extracts line feature information from the images, and compares it with a normal line feature template in real time. When the deviation between the feature information of a certain line location and the normal line feature template exceeds a preset allowable threshold, and the location is not included in the abnormal point location information identified in step 10, the location is determined to be a newly added abnormal point. At the same time, the positioning module of the inspection device synchronously acquires the location information of the newly added abnormal point.

[0081] Optionally, the inspection device updates the target inspection path based on the acquired location information and distribution relationship of the newly added anomaly points. For example, it analyzes the spatial proximity and physical connection relationship between the newly added anomaly points and the original local anomaly clusters, and determines whether the newly added anomaly points are extensions of the original local anomaly clusters or whether a new local anomaly cluster needs to be formed. Based on the determination results, the direction of the original inspection path is adjusted to include the newly added anomaly points in the inspection coverage area, while ensuring that the updated inspection path still meets the shortest path requirement, thus obtaining the updated inspection path, as detailed in steps 401 to 404.

[0082] Furthermore, the inspection device continues to inspect along the updated inspection path, sequentially completing the inspection of existing abnormal points, newly added abnormal points, and other line sections along the path, until all abnormal points and the line along the path are inspected.

[0083] The embodiments of the present invention solve the problems of easily overlooking dynamically occurring anomalies, delaying processing opportunities, and increasing inspection time. They enable the inspection path to dynamically adapt to the actual line status during the inspection process, realize the rational allocation of inspection resources, improve inspection efficiency, and dynamically cover abnormal points, thus ensuring the dynamic and efficient operation of inspection.

[0084] Optionally, the processes of steps 201 to 204 include:

[0085] Step 201: For any two different first line anomaly points and second line anomaly points, if there is at least one relationship between the first line anomaly point and the second line anomaly point, either a spatial proximity relationship or a physical connection relationship, then the first line anomaly point and the second line anomaly point are determined as an anomaly point pair.

[0086] Optionally, for any two different first line abnormal points and second line abnormal points, the inspection device judges the relationship between each group of first line abnormal points and second line abnormal points. If there is at least one relationship between the two, namely, spatial proximity relationship and line physical connection relationship (i.e., there is a spatial proximity relationship, a line physical connection relationship, or both relationships exist simultaneously), then the two abnormal points in the group are determined as an abnormal point pair; if neither of the two satisfies the above two relationships, then they do not constitute an abnormal point pair.

[0087] In one embodiment, the inspection robot has identified three abnormal points on the power line: abnormal point 1 (3D coordinates: X1, Y1, Z1), abnormal point 2 (3D coordinates: X2, Y2, Z2), and abnormal point 3 (3D coordinates: X3, Y3, Z3). The relationships between the abnormal points have been analyzed: the actual straight-line distance between abnormal point 1 and abnormal point 2 is 8 meters (less than the preset proximity distance threshold of 10 meters), indicating a spatial proximity relationship. Furthermore, the two points are located on the same continuous overhead transmission line, indicating a physical connection. The actual straight-line distance between abnormal point 1 and abnormal point 3 is 15 meters (greater than the preset proximity distance threshold of 10 meters), and the two points are located on two unconnected branches, indicating no physical connection. The actual straight-line distance between abnormal point 2 and abnormal point 3 is 12 meters (greater than the preset proximity distance threshold of 10 meters), and the two points are located on two unconnected branches, indicating no physical connection. The inspection device performs pairwise combinations on the three abnormal points, resulting in three pairs: abnormal point 1 and abnormal point 2, abnormal point 1 and abnormal point 3, and abnormal point 2 and abnormal point 3. For each pair, a relationship judgment is made: abnormal point 1 and abnormal point 2 satisfy the condition of "at least one relationship exists," and are therefore identified as an abnormal point pair; abnormal point 1 and abnormal point 3, and abnormal point 2 and abnormal point 3 do not satisfy this condition and do not constitute an abnormal point pair, resulting in one abnormal point pair: (abnormal point 1, abnormal point 2).

[0088] Step 202: Merge pairs of anomalies that share the same line anomaly point to obtain adjacent anomaly point groups. Any two line anomalies within an adjacent anomaly point group can be connected to each other through the path in the anomaly point pair.

[0089] Optionally, the inspection device determines whether there is a shared line anomaly point between different pairs of anomalies (i.e., the same line anomaly point exists simultaneously in two or more different pairs of anomalies). If a shared line anomaly point exists, all pairs of anomalies containing the shared anomaly point are merged, and the resulting set is the adjacent anomaly point group. If no shared line anomaly point exists, each pair of anomalies is treated as a separate adjacent anomaly point group. Further, the inspection device verifies the merged adjacent anomaly point groups to ensure that any two line anomalies within a group can be interconnected through paths within the pairs of anomalies in the group. Interconnectivity of paths indicates that starting from one line anomaly point, one can reach any other line anomaly point within the group by progressively associating with consecutive pairs of anomalies.

[0090] In one embodiment, step 201 obtains one pair of abnormal points (abnormal point 1, abnormal point 2). The inspection device subsequently identifies abnormal point 4 (3D coordinates: X4, Y4, Z4). After analysis in step 201, it is determined that there is a physical connection between abnormal point 2 and abnormal point 4 (the relevant line sections are directly connected via connectors), thus obtaining another pair of abnormal points (abnormal point 2, abnormal point 4). At this point, step 201 obtains a total of two pairs of abnormal points: (abnormal point 1, abnormal point 2) and (abnormal point 2, abnormal point 4).

[0091] Furthermore, the inspection device identifies a shared anomaly point between the two pairs of anomaly points: both pairs contain anomaly point 2, meaning anomaly point 2 is a shared line anomaly point. Therefore, the two pairs of anomaly points are merged to obtain an adjacent anomaly point group, which contains the following line anomalies: anomaly point 1, anomaly point 2, and anomaly point 4. Connectivity verification is then performed on this group: anomaly point 1 can be connected to anomaly point 2 via the anomaly point pair (anomaly point 1, anomaly point 2), and anomaly point 2 can be connected to anomaly point 4 via the anomaly point pair (anomaly point 2, anomaly point 4). Therefore, anomaly point 1 can be connected to anomaly point 4 via anomaly point 2, and any two anomalies within the group are interconnected, thus the verification passes.

[0092] Step 203: For each adjacent abnormal point group, construct an abnormal point connected subgraph with each abnormal point in the group as a node and the physical connection relationship or spatial proximity relationship between each abnormal point in the group as an edge.

[0093] Optionally, for each adjacent anomaly point group, the inspection device defines each line anomaly point within the group as a node (a graphical identifier representing the line anomaly point, whose attributes include the location information of the anomaly point); and defines the physical connection relationship or spatial proximity relationship between any two line anomaly points within the group as an edge (a graphical connection representing the relationship between two nodes, whose attributes include the type of relationship, i.e., physical connection relationship or spatial proximity relationship). Further, based on nodes and edges, the inspection device uses graphical construction logic to connect the nodes within each adjacent anomaly point group through corresponding edges, obtaining an anomaly point connected subgraph. Therefore, the anomaly point connected subgraph refers to a graphical structure that can intuitively reflect all line anomaly points and their relationships within an adjacent anomaly point group.

[0094] In one embodiment, the adjacent anomaly point group includes anomaly point 1, anomaly point 2, and anomaly point 4. For this adjacent anomaly point group, anomaly point 1, anomaly point 2, and anomaly point 4 are respectively identified as node 1, node 2, and node 4. Node 1 and node 2 have a physical connection relationship and a spatial proximity relationship, and an edge (edge ​​1) containing both relationship attributes is determined. Node 2 and node 4 have a physical connection relationship, and an edge (edge ​​2) containing the physical connection relationship attribute is determined. Further, a graphical construction logic is used to construct a connected subgraph of the anomaly points: taking node 1, node 2, and node 4 as three independent nodes, node 1 is connected to node 2 through edge 1, and node 2 is connected to node 4 through edge 2, resulting in a connected subgraph of "node 1—edge 1—node 2—edge 2—node 4". In this subgraph, node 1 can be connected to node 4 through edges 1 and 2, clearly reflecting the association relationship of the three anomaly points within the adjacent anomaly point group.

[0095] Step 204: Based on the route indicated by the physical connection relationship of the route in the connected subgraph of each anomaly point, construct the distribution relationship of the anomaly points.

[0096] Optionally, the inspection device constructs the distribution relationship of abnormal points based on the route indicated by the physical connection relationship of the line in the connection subgraph of each abnormal point, as in steps 2041 to 2044.

[0097] The embodiments of the present invention accurately construct the distribution relationship from scattered anomalies to structured anomalies, intuitively presenting the association types and connection paths between anomalies, and realizing the efficiency and accuracy of inspection path decision-making.

[0098] Optionally, the process of steps 2041 to 2044 includes:

[0099] Step 2041: Based on the route indicated by the physical connection relationship in the connectivity subgraph of each anomaly point, connect adjacent anomaly points sequentially to obtain anomaly point sequence paths. There is a physical connection relationship between adjacent anomaly points in each anomaly point sequence path.

[0100] Optionally, the inspection device analyzes the connected subgraph of each abnormal point to determine the abnormal point of the line corresponding to each node in the subgraph, the relationship type (physical connection relationship or spatial proximity relationship) corresponding to each edge, and the line direction indicated by the physical connection relationship (i.e. the actual extension direction of the line to be inspected, such as the extension direction from the starting point to the end point of the line, the branching direction from the main line to the branch line, etc.).

[0101] Furthermore, for each anomaly point connected subgraph, the inspection device filters out all edges whose corresponding relationship type is physical line connection, excluding edges that only have spatial proximity. Next, based on the line direction indicated by the physical line connection relationship, starting from any anomaly point in the subgraph, it sequentially connects adjacent anomaly points with which it has a physical line connection relationship, obtaining a preliminary point list. Finally, the preliminary point list is verified to ensure that there is a physical line connection relationship between adjacent anomaly points in the point list, and that the point list completely covers all anomaly points with physical line connection relationships in the subgraph. After verification, the point list is determined as the anomaly point sequence path. Therefore, the anomaly point sequence path refers to the path formed by the orderly connection of anomaly points according to the physical line direction, reflecting the actual distribution order of anomaly points on the line to be inspected. If an anomaly point connected subgraph contains multiple independent physical line connection branches, corresponding anomaly point sequence paths are constructed for each branch.

[0102] In one embodiment, the abnormal point connected subgraph constructed in step 203 includes three nodes (node ​​1 corresponds to abnormal point 1, node 2 corresponds to abnormal point 2, and node 4 corresponds to abnormal point 4), two edges (edge ​​1 corresponds to the physical connection relationship and spatial proximity relationship between node 1 and node 2, and edge 2 corresponds to the physical connection relationship between node 2 and node 4), and the physical topology of the line records that the line direction in this area extends from abnormal point 1 to abnormal point 4.

[0103] Further, after the inspection device extracts the connected subgraph of the abnormal point, it filters out edge 1 and edge 2 (both of which are edges corresponding to the physical connection relationship of the line) and excludes edges that only have spatial proximity relationship; according to the line direction (from abnormal point 1 to abnormal point 4), taking abnormal point 1 corresponding to node 1 as the starting node, it connects abnormal point 2 corresponding to node 2 through edge 1, and then connects abnormal point 4 corresponding to node 4 through edge 2, to obtain a preliminary point sequence: abnormal point 1—abnormal point 2—abnormal point 4; the point sequence is verified to confirm that there is a physical connection relationship between adjacent abnormal point 1 and abnormal point 2, and between abnormal point 2 and abnormal point 4, and it completely covers all abnormal points in the subgraph that have a physical connection relationship of the line. After the verification is passed, the point sequence is determined to be the abnormal point sequence path.

[0104] Step 2042: For any target path in the anomaly point sequence path, if the number of line anomalies it contains is greater than or equal to the preset minimum number of anomalies threshold, then the line segment covered by the target path is determined as a local anomaly cluster area.

[0105] Optionally, the preset minimum number of abnormal points threshold is the number of critical abnormal points used to determine whether the line segment corresponding to the abnormal point sequence path constitutes a local abnormal cluster area. It is determined based on factors such as the type of line to be inspected, the inspection accuracy requirements, and the statistical results of historical abnormal data, such as setting it to 3.

[0106] Optionally, the inspection device determines the number of line anomalies contained in the anomaly sequence path.

[0107] Furthermore, for each abnormal point sequence path (target path), if the number of abnormal points in the target path is greater than or equal to the preset minimum abnormal point threshold, it is determined that the line segment corresponding to the target path (i.e., the actual section of the line to be inspected between the starting and ending abnormal points in the target path) has a concentrated abnormality, and the inspection device identifies the line segment as a local abnormality cluster area; if the number of abnormal points in the target path is less than the preset minimum abnormal point threshold, it is determined that the line segment corresponding to the path has no concentrated abnormality and does not constitute a local abnormality cluster area.

[0108] In one embodiment, in addition to obtaining the abnormal point sequence path 1 (abnormal point 1—abnormal point 2—abnormal point 4, containing 3 line abnormal points) in step 2041, an abnormal point sequence path 2 (abnormal point 5—abnormal point 6, containing 2 line abnormal points) is also obtained through another abnormal point connected subgraph. The preset minimum abnormal point count threshold is 3. The inspection device extracts two abnormal point sequence paths and judges them as target paths respectively: the number of line abnormal points contained in target path 1 (abnormal point 1—abnormal point 2—abnormal point 4) is 3, which is equal to the preset minimum abnormal point count threshold. Therefore, the line segment corresponding to this path (the overhead transmission line segment between abnormal point 1 and abnormal point 4) is determined as a local abnormal cluster area; the number of line abnormal points contained in target path 2 (abnormal point 5—abnormal point 6) is 2, which is less than the preset minimum abnormal point count threshold. Therefore, the line segment corresponding to this path (the overhead transmission line segment between abnormal point 5 and abnormal point 6) does not constitute a local abnormal cluster area.

[0109] Step 2043: For each local anomaly cluster, the line anomaly points corresponding to its starting and ending ends are used as the boundary points of the anomaly region. Combining the connectivity relationships between the various local anomaly clusters in the anomaly point connectivity subgraph, an anomaly distribution topology is constructed. The anomaly distribution topology represents the interconnectivity of each local anomaly cluster on the line to be inspected.

[0110] Optionally, for each local anomaly cluster area, the inspection device locates the corresponding anomaly point sequence path, determines the line anomaly point corresponding to the beginning of the path as the boundary point of the beginning anomaly area, and determines the line anomaly point corresponding to the end of the path as the boundary point of the end anomaly area (i.e., the edge anomaly point that can define the range of the local anomaly cluster area, through which the beginning and end positions of the local anomaly cluster area can be clearly identified).

[0111] Furthermore, the inspection device extracts the connected subgraph of abnormal points and the preset physical topology of the line, and sorts out the connection relationship between different local abnormal clusters (i.e., whether the line segments corresponding to two local abnormal clusters are directly connected through normal sections of the line to be inspected, indirectly connected through other local abnormal clusters, or have no connection relationship at all). Finally, using topology structured logic, each local abnormal cluster is used as an independent topology node, and the connection relationship between each local abnormal cluster is used as the topology edge. Combined with the abnormal area boundary point information corresponding to each topology node, an abnormal distribution topology is constructed. Therefore, the abnormal distribution topology refers to a structured model that can intuitively represent the mutual connection relationship of each local abnormal cluster on the line to be inspected and the boundary range of each cluster.

[0112] In one embodiment, step 2042 identifies local anomaly cluster 1 (corresponding to anomaly point sequence path 1: anomaly point 1—anomaly point 2—anomaly point 4). If local anomaly cluster 2 is also identified (corresponding to anomaly point sequence path 3: anomaly point 7—anomaly point 8—anomaly point 9, containing 3 line anomalies), then the line segment corresponding to local anomaly cluster 1 and the line segment corresponding to local anomaly cluster 2 are directly connected by a normal overhead transmission line. Furthermore, the boundary point of the anomaly area at the end of local anomaly cluster 1 is anomaly point 4, and the boundary point of the anomaly area at the beginning of local anomaly cluster 2 is anomaly point 7.

[0113] The boundary points of the two local anomaly clusters are as follows: the starting boundary point of local anomaly cluster 1 is anomaly point 1, and the ending boundary point is anomaly point 4; the starting boundary point of local anomaly cluster 2 is anomaly point 7, and the ending boundary point is anomaly point 9. Using local anomaly cluster 1 and local anomaly cluster 2 as two topological nodes, and the direct connection between them as topological edges, and combining the boundary points of the anomaly regions corresponding to each topological node, an anomaly distribution topology is constructed as follows: Topological node 1 (local anomaly cluster 1, boundary points: anomaly point 1, anomaly point 4) — topological edge (direct connection) — topological node 2 (local anomaly cluster 2, boundary points: anomaly point 7, anomaly point 9).

[0114] Step 2044: Integrate the local anomaly clusters, anomaly region boundary points and their interconnections contained in the anomaly distribution topology to obtain the anomaly point distribution relationship.

[0115] Optionally, the inspection device classifies and organizes all local anomaly clusters in the topology, the boundary points of the anomaly areas corresponding to each local anomaly cluster, and the interconnections between the local anomaly clusters. It removes duplicate information (such as duplicate connection relationships and boundary point information) and supplements detailed data such as the line segment location information corresponding to each local anomaly cluster and the spatial coordinate information of the boundary points of each anomaly area. Finally, the organized information is integrated to obtain the distribution relationship of anomaly points.

[0116] In one embodiment, the abnormal distribution topology constructed in step 2043 includes two local abnormal clusters, four abnormal region boundary points, and a set of direct connections. After the inspection device extracts this topology, it categorizes and organizes the information: local abnormal cluster information (local abnormal cluster 1: the corresponding line segment is the overhead transmission line between abnormal point 1 and abnormal point 4, with spatial coordinates XXX; local abnormal cluster 2: the corresponding line segment is the overhead transmission line between abnormal point 7 and abnormal point 9, with spatial coordinates YYY), abnormal region boundary point information (abnormal point 1: spatial coordinates XXX1; abnormal point 4: spatial coordinates XXX4; abnormal point 7: spatial coordinates YYY1; abnormal point 9: spatial coordinates YYY4), and interconnection information (local abnormal cluster 1 and local abnormal cluster 2 are directly connected through the normal line segment between abnormal point 4 and abnormal point 7).

[0117] This invention enables the accurate transformation of anomaly distribution information from scattered to structured, accurately identifying local anomaly clusters and boundaries, clarifying the connection relationships within these clusters, and ensuring the accuracy of inspection path decisions based on anomaly distribution relationships. This allows the final target inspection path to closely match the actual distribution and relationships of anomalies, avoiding repeated backtracking between scattered anomalies and excessively long ineffective travel distances. It also reduces wasted time in areas without anomalies, solving the problems of wasted inspection resources and low efficiency in anomaly detection.

[0118] Optionally, the processes of steps 301 to 304 include:

[0119] Step 301: Based on the distribution relationship of abnormal points, the local abnormal cluster area that is directly connected to the first current inspection position through the physical connection relationship of the line of the abnormal area boundary point is determined as the initial reachable abnormal area.

[0120] Optionally, for each local anomaly cluster in the distribution relationship of anomalies, the inspection device verifies whether there is a physical connection between the boundary point of its anomaly area and the first current inspection position, and whether it is a direct connection. Direct connection means that the two positions can be reached directly through the physical section of the line to be inspected without passing through other local anomaly clusters, and there are no line interruptions, obstacles or other impassable situations in the path. This connectivity information can be queried and confirmed from the preset physical topology of the line.

[0121] Furthermore, the inspection device identifies local anomaly clusters that are physically connected to the first current inspection location and are directly connected as the initial reachable anomaly area.

[0122] In one embodiment, the distribution of abnormal points includes two local abnormal clusters: Local Abnormal Cluster 1 (boundary points: abnormal point 1, abnormal point 4) and Local Abnormal Cluster 2 (boundary points: abnormal point 7, abnormal point 9). These two clusters are directly connected via a normal route section between abnormal point 4 and abnormal point 7. If the first current inspection position is located 5 meters from abnormal point 1 on the route inspection track, the preset route physical topology shows a direct physical connection between this first current inspection position and abnormal point 1 (abnormal point 1 can be reached directly via the inspection track and a side passage), with no obstructions. The inspection device verifies the two local abnormal clusters: Abnormal point 1, the boundary point of Local Abnormal Cluster 1, is directly connected to the first current inspection position, satisfying the judgment condition; Abnormal points 7 and 9, the boundary points of Local Abnormal Cluster 2, must pass through Local Abnormal Cluster 1 to reach the first current inspection position, thus not satisfying the direct connection condition.

[0123] Therefore, the local anomaly cluster 1 is determined as the initial reachable anomaly region.

[0124] Step 302: Starting from the initially reachable abnormal region, based on the physical connection relationship of the line indicated by the abnormal distribution topology in the abnormal point distribution relationship, traverse the remaining local abnormal clusters directly connected to it in sequence to obtain the abnormal region adjacency sequence.

[0125] Optionally, the inspection device uses the initially reachable abnormal region as the starting point for traversal. Based on the topological edges of the abnormal distribution topology in the abnormal point distribution relationship, it filters out all remaining local abnormal clusters that have a direct connection relationship with the initially reachable abnormal region (i.e., other local abnormal clusters in the abnormal point distribution relationship besides the initially reachable abnormal region). Furthermore, this embodiment of the invention adopts breadth-first traversal logic, that is, it first traverses all regions directly connected to the starting point, and then traverses the directly connected regions of each connected region in turn. Starting from the initially reachable abnormal region, it visits the filtered local abnormal clusters in turn and records the name of the region visited each time. Finally, it arranges the initially reachable abnormal region and all local abnormal clusters visited during the traversal in the order of visit to obtain the abnormal region adjacency sequence.

[0126] In one embodiment, the anomaly distribution topology is: Local anomaly cluster 1 (initial reachable anomaly area) — direct connection — Local anomaly cluster 2. If a new local anomaly cluster 3 (boundary points: anomaly point 10, anomaly point 12) is added to the anomaly point distribution relationship, and is directly connected to local anomaly cluster 2 through the normal line segment between anomaly point 9 and anomaly point 10, the anomaly distribution topology is updated to: Local anomaly cluster 1 — direct connection — Local anomaly cluster 2 — direct connection — Local anomaly cluster 3.

[0127] The inspection device takes the local anomaly cluster area 1 as the starting point for traversal and adopts a breadth-first traversal logic: it visits the starting local anomaly cluster area 1, then selects and visits the local anomaly cluster area 2 that is directly connected to it, and then selects and visits the remaining local anomaly cluster area 3 that is directly connected to local anomaly cluster area 2; it records the area names according to the access order to obtain the anomaly area adjacency sequence: local anomaly cluster area 1, local anomaly cluster area 2, local anomaly cluster area 3.

[0128] Step 303: Sort each local abnormal cluster in the abnormal region adjacency sequence according to its physical arrangement on the transmission line to obtain the abnormal region access order that extends unidirectionally from the initially reachable abnormal region along the physical connection relationship of the line.

[0129] Optionally, the inspection device determines the physical arrangement order of the lines to be inspected recorded in the preset physical topology of the lines (i.e., the natural extension order of the lines to be inspected from the starting point to the end point, the branching order of the main line and the branch line, and other actual physical layout order, such as the ascending order of the tower numbers of the transmission line, the directional order of the line direction, etc.); secondly, it clarifies the sorting rules: taking the initially accessible abnormal area as the starting point, according to the physical arrangement order of each local abnormal cluster area on the line to be inspected, all local abnormal cluster areas in the adjacent sequence of the abnormal area are adjusted to a unidirectional extension order (i.e., the sorted order is consistent with the actual extension direction of the line, avoiding back-and-forth arrangements and ensuring the continuity of the inspection path).

[0130] Furthermore, the inspection device adjusts the adjacent sequence of abnormal areas according to the sorting rules to obtain the access order of abnormal areas. Therefore, the access order of abnormal areas refers to the order of local abnormal clusters arranged sequentially along the physical extension direction of the line, starting from the initially reachable abnormal area.

[0131] In one embodiment, the abnormal region adjacency sequence obtained in step 302 is: local abnormal cluster area 1, local abnormal cluster area 2, and local abnormal cluster area 3. The preset line physical topology shows that the physical arrangement order of the transmission line to be inspected is a unidirectional extension from local abnormal cluster area 1 (corresponding to the start end of the line segment) to local abnormal cluster area 3 (corresponding to the end end of the line segment). The specific order is: the line segment where local abnormal cluster area 1 is located—the line segment where local abnormal cluster area 2 is located—the line segment where local abnormal cluster area 3 is located, which conforms to the increasing direction of the tower number.

[0132] The inspection device verifies the adjacency sequence of abnormal areas according to the physical arrangement of the lines. It finds that the order of the adjacency sequence is consistent with the physical extension order of the lines and requires no adjustment. Therefore, the adjacency sequence of the abnormal areas is directly determined as the access order for the abnormal areas: Local Abnormal Cluster Area 1, Local Abnormal Cluster Area 2, Local Abnormal Cluster Area 3. If the adjacency sequence is reversed (e.g., Local Abnormal Cluster Area 3, Local Abnormal Cluster Area 2, Local Abnormal Cluster Area 1), it is adjusted to the order consistent with the physical arrangement of the lines.

[0133] Step 304: Based on the boundary points of the abnormal regions of two adjacent local abnormal clusters in the abnormal region access sequence, the inspection path is determined to identify the target inspection path.

[0134] Optionally, the inspection device makes inspection path decisions based on the boundary points of two adjacent local abnormal clusters in the abnormal area access sequence, and determines the target inspection path, as in steps 3041 to 3044.

[0135] The target inspection path generated by the embodiments of the present invention can accurately cover all local anomaly clusters, taking into account both path length and inspection efficiency, while conforming to the actual physical layout of the line to be inspected. It solves the problems of incomplete coverage, path redundancy, and disconnection from the actual line scenario in traditional inspection path planning, improves the targeting and work efficiency of line inspection, and ensures dynamic and efficient inspection operations.

[0136] Optionally, the processes of steps 3041 to 3044 include:

[0137] Step 3041: For adjacent local abnormal clusters in the abnormal region access sequence, determine the inter-regional connection path segment based on the physical connection relationship between the abnormal region boundary point corresponding to the termination end of the previous local abnormal cluster and the abnormal region boundary point corresponding to the starting end of the next local abnormal cluster.

[0138] Optionally, for any two adjacent local anomaly clusters in the anomaly area access sequence, the inspection device determines the correspondence between the preceding and following local anomaly clusters, locating the boundary point of the anomaly region at the end of the preceding local anomaly cluster (i.e., the anomaly point at the end of the anomaly point sequence path corresponding to the local anomaly cluster) and the boundary point of the anomaly region at the beginning of the following local anomaly cluster (i.e., the anomaly point at the beginning of the anomaly point sequence path corresponding to the local anomaly cluster). Further, based on the physical connection relationship between the two boundary points determined by the physical topology of the line, the continuous line segment between them is confirmed (i.e., the section of the line to be inspected and its associated inspection channel that is uninterrupted between the two boundary points and can ensure the smooth passage of the inspection device). Finally, the inspection device determines this continuous line segment as an inter-regional connection path segment, which is used to connect the path segments of two adjacent local anomaly clusters, achieving a smooth transition from one local anomaly cluster to the next.

[0139] In one embodiment, the abnormal region access order obtained in step 303 is: local abnormal cluster area 1, local abnormal cluster area 2, and local abnormal cluster area 3; the boundary point information of each local abnormal cluster area is: local abnormal cluster area 1 (starting boundary point: abnormal point 1, ending boundary point: abnormal point 4), local abnormal cluster area 2 (starting boundary point: abnormal point 7, ending boundary point: abnormal point 9), and local abnormal cluster area 3 (starting boundary point: abnormal point 10, ending boundary point: abnormal point 12); the abnormal distribution topology shows that local abnormal cluster area 1 and local abnormal cluster area 2 are directly connected through the normal line segment between abnormal point 4 and abnormal point 7, and local abnormal cluster area 2 and local abnormal cluster area 3 are directly connected through the normal line segment between abnormal point 9 and abnormal point 10.

[0140] For local anomaly cluster 1 and local anomaly cluster 2, locate the termination boundary point (anomaly point 4) of the former region (local anomaly cluster 1) and the starting boundary point (anomaly point 7) of the latter region (local anomaly cluster 2), query the physical topology of the line to confirm that there is a continuous normal line segment between the two, and determine the segment as inter-regional connection path segment 1; for local anomaly cluster 2 and local anomaly cluster 3, locate the termination boundary point (anomaly point 9) of the former region (local anomaly cluster 2) and the starting boundary point (anomaly point 10) of the latter region (local anomaly cluster 3), confirm the continuous normal line segment between the two, and determine the segment as inter-regional connection path segment 2.

[0141] Step 3042: Based on the physical connection relationship between the first current inspection position and the boundary point of the abnormal area corresponding to the starting end of the initially reachable abnormal area, determine the continuous line segment from the first current inspection position to the boundary point of the abnormal area corresponding to the starting end of the initially reachable abnormal area, and obtain the initial access path segment.

[0142] Optionally, based on the physical connection relationship between the first current inspection location and the boundary point of the initial accessible abnormal area, as confirmed by the physical topology of the line, the inspection device filters out a continuous line segment (i.e., a line and auxiliary passage segment without line interruption and meeting the inspection device's access requirements) that can ensure the smooth passage of the inspection device. Finally, the inspection device determines this continuous line segment as the initial access path segment. Therefore, the initial access path segment refers to the path segment connecting the current location of the inspection device and the starting point of the abnormal area inspection path, which is the initial connection path for the inspection device to enter the abnormal area for inspection.

[0143] In one embodiment, the initial reachable abnormal area determined in step 301 is a local abnormality cluster area 1, and its initial boundary point is abnormal point 1. The first current inspection position of the inspection device is the inspection track position 5 meters away from abnormal point 1. The preset physical topology of the line shows that there is a continuous inspection channel segment between the first current inspection position and abnormal point 1. This channel is unobstructed, and its width, load-bearing capacity, and other parameters meet the passage requirements of the inspection device. After the inspection device queries and confirms the above connection relationship and continuous channel segment, it determines the continuous inspection channel segment from the first current inspection position to abnormal point 1 as the initial access path segment, ensuring that the inspection device can smoothly move from its current position to the initial boundary point of the local abnormality cluster area 1 and complete the initial access of the inspection path.

[0144] Step 3043: For each local anomaly cluster in the anomaly region access sequence, generate bidirectional candidate traversal paths with opposite traversal directions based on its corresponding anomaly point sequence path. The first candidate path traverses from the anomaly region boundary point at the starting end to the anomaly region boundary point at the ending end, and the second candidate path traverses from the anomaly region boundary point at the ending end to the anomaly region boundary point at the starting end.

[0145] Optionally, for each local anomaly cluster in the anomaly area access sequence, the inspection device generates two candidate traversal paths with opposite directions based on its anomaly point sequence path: the first candidate path is a forward traversal path, traversing from the starting boundary point of the local anomaly cluster to the ending boundary point of the local anomaly cluster, with the path order consistent with the natural order of the anomaly point sequence path; the second candidate path is a reverse traversal path, traversing from the ending boundary point of the local anomaly cluster to the starting boundary point of the local anomaly cluster, with the path order reversed from the natural order of the anomaly point sequence path; finally, the two generated candidate traversal paths with opposite directions are stored in the database as the basis path segments for subsequent path decisions. Both candidate paths must completely cover all anomaly point sequence path segments within the local anomaly cluster to ensure comprehensive coverage of anomalies within the cluster.

[0146] In one embodiment, the abnormal point sequence path corresponding to the local abnormal cluster area 1 is: abnormal point 1—abnormal point 2—abnormal point 4 (the natural order is from the start end to the end end); the abnormal point sequence path corresponding to the local abnormal cluster area 2 is: abnormal point 7—abnormal point 8—abnormal point 9; and the abnormal point sequence path corresponding to the local abnormal cluster area 3 is: abnormal point 10—abnormal point 11—abnormal point 12. For each cluster, generate bidirectional candidate traversal paths: For local abnormal cluster 1, the first candidate path (forward) is abnormal point 1—abnormal point 2—abnormal point 4, and the second candidate path (reverse) is abnormal point 4—abnormal point 2—abnormal point 1; For local abnormal cluster 2, the first candidate path (forward) is abnormal point 7—abnormal point 8—abnormal point 9, and the second candidate path (reverse) is abnormal point 9—abnormal point 8—abnormal point 7; For local abnormal cluster 3, the first candidate path (forward) is abnormal point 10—abnormal point 11—abnormal point 12, and the second candidate path (reverse) is abnormal point 12—abnormal point 11—abnormal point 10.

[0147] Step 3044: Based on the initial access path segment, the inter-regional connection path segments of adjacent local anomaly clusters, and the bidirectional candidate traversal path of each local anomaly cluster, the inspection path decision is made to determine the target inspection path.

[0148] Optionally, the inspection device makes inspection path decisions based on the initial access path segment, the inter-regional connection path segment of adjacent local anomaly clusters, and the bidirectional candidate traversal path of each local anomaly cluster, and determines the target inspection path, as in steps 30441 to 30443.

[0149] The target inspection path obtained by the final decision in the embodiments of the present invention can fully cover all local anomaly cluster areas, taking into account both the shortest path length and the maximum inspection efficiency. At the same time, it conforms to the physical layout of the line to be inspected and the access requirements of the inspection device, solving the problems of incomplete coverage and path redundancy, improving the efficiency of inspection of abnormal areas of the line, and thus ensuring the dynamic efficiency of inspection operations.

[0150] Optionally, the processes of steps 30441 to 30443 include:

[0151] Step 30441: Match the endpoint of the initial access path segment with the two boundary points of the two abnormal regions of the initially reachable abnormal region. If the endpoint of the initial access path segment is connected to the boundary point of the abnormal region at the beginning of the initially reachable abnormal region through a physical connection, then the candidate traversal path in the traversal direction from the beginning to the end is determined as the target traversal path. If the endpoint of the initial access path segment is connected to the boundary point of the abnormal region at the end of the initially reachable abnormal region, then the candidate traversal path in the traversal direction from the end to the beginning is determined as the target traversal path.

[0152] Optionally, the inspection device performs connectivity matching between the endpoint of the initial access path segment and the two boundary points (starting end and ending end) of the initially reachable abnormal area. The connectivity judgment is based on the physical connection relationship of the line (i.e., the connection relationship between the two locations that can be directly reached through the physical segment of the line to be inspected, without interruption or impassable obstacles, which can be confirmed by querying the physical topology of the line). Further, the target traversal path is determined according to the matching results: if the endpoint of the initial access path segment is connected to the boundary point of the starting abnormal area of ​​the initially reachable abnormal area through the physical connection relationship of the line, it means that the starting end can be directly entered from the initial access path, and the candidate traversal path in the traversal direction from the starting end to the ending end of the area is determined as the target traversal path; if the endpoint of the initial access path segment is connected to the boundary point of the ending abnormal area of ​​the initially reachable abnormal area through the physical connection relationship of the line, the candidate traversal path in the traversal direction from the ending end to the starting end is determined as the target traversal path.

[0153] In one embodiment, the starting point of the initial access path segment obtained in step 3042 is the first current inspection position of the inspection device (the inspection track position 5 meters away from anomaly point 1), and the ending point is anomaly point 1; the initial reachable anomaly area determined in step 301 is the local anomaly cluster area 1, and its two anomaly area boundary points are the starting end boundary point (anomaly point 1) and the ending end boundary point (anomaly point 4), respectively; the physical topology of the line shows that the ending point (anomaly point 1) of the initial access path segment and the starting end boundary point (anomaly point 1) are at the same location, and are naturally connected through the physical connection relationship of the line.

[0154] Furthermore, the matching judgment is as follows: the endpoint of the initial access path segment is connected to the starting boundary point (abnormal point 1) of the local abnormal cluster area 1. Therefore, the candidate traversal path (abnormal point 1 - abnormal point 2 - abnormal point 4) in the traversal direction from the starting end to the ending end of the area is determined as the target traversal path of the local abnormal cluster area 1, ensuring that the initial access path and the traversal path in the area can be directly and smoothly connected.

[0155] Step 30442: For each local anomaly cluster starting from the second local anomaly cluster in the anomaly region access sequence, compare the endpoint of the candidate traversal path of the previous local anomaly cluster with the two anomaly region boundary points of the current local anomaly cluster. If the endpoint is connected to the anomaly region boundary point at the start of the current local anomaly cluster through a physical connection, the candidate traversal path in the traversal direction from the start to the end is determined as the target traversal path. If the endpoint is connected to the anomaly region boundary point at the end, the candidate traversal path in the traversal direction from the end to the start is determined as the target traversal path. By comparing the endpoint of the candidate traversal path with the two anomaly region boundary points, the target traversal paths for the remaining local anomaly clusters are obtained.

[0156] Optionally, the inspection device starts from the second local anomaly cluster in the abnormal area access sequence and processes each local anomaly cluster (the current local anomaly cluster) sequentially. Next, it executes matching logic: comparing the endpoint of the target traversal path of the previous local anomaly cluster (the preceding local anomaly cluster) with the two anomaly region boundary points (start and end) of the current local anomaly cluster, while simultaneously querying the physical topology of the line to determine whether the endpoint of the preceding path is connected to the two boundary points of the current region through a physical connection. Finally, it determines the target traversal path of the current region based on the matching results: if the endpoint of the target traversal path of the preceding local anomaly cluster is connected to the starting point of the current local anomaly cluster, the candidate traversal path from the starting point to the ending point of the current region is determined as the target traversal path; if the endpoint of the preceding path is connected to the ending point of the current region, the candidate traversal path from the ending point to the starting point of the current region is determined as the target traversal path. This logic is applied sequentially to all remaining local anomaly clusters to obtain the target traversal path for each region.

[0157] In one embodiment, the abnormal region access order obtained in step 303 is: local abnormal cluster region 1, local abnormal cluster region 2, and local abnormal cluster region 3; the target traversal path of local abnormal cluster region 1 determined in step 30441 is abnormal point 1—abnormal point 2—abnormal point 4, and the path endpoint is abnormal point 4; information of each remaining region: local abnormal cluster region 2 (starting boundary point: abnormal point 7, ending boundary point: abnormal point 9, bidirectional candidate traversal path: forward is abnormal point 7—abnormal point 8—abnormal point 9, reverse is abnormal point 9—abnormal point 8—abnormal point 7), local abnormal cluster region 3 (starting boundary point: abnormal point 10, ending boundary point: abnormal point 1) 2. Bidirectional candidate traversal path: the forward direction is anomaly point 10—anomaly point 11—anomaly point 12, and the reverse direction is anomaly point 12—anomaly point 11—anomaly point 10); the physical topology of the line shows that the endpoint of the target traversal path of local anomaly cluster 1 (anomaly point 4) is connected to the starting boundary point of local anomaly cluster 2 (anomaly point 7) through inter-regional connection path segment 1 (the normal line segment between anomaly point 4 and anomaly point 7), and the endpoint of the target traversal path of local anomaly cluster 2 (if it is anomaly point 9) is connected to the starting boundary point of local anomaly cluster 3 (anomaly point 10) through inter-regional connection path segment 2 (the normal line segment between anomaly point 9 and anomaly point 10).

[0158] Further, the remaining regions are processed sequentially: Local anomaly cluster region 2 (the second region) is processed by comparing the endpoint of the target traversal path (anomaly point 4) of the preceding region (local anomaly cluster region 1) with the two boundary points of local anomaly cluster region 2. It is confirmed that anomaly point 4 is connected to the starting boundary point (anomaly point 7). Therefore, the forward candidate traversal path of local anomaly cluster region 2 (anomaly point 7—anomaly point 8—anomaly point 9) is determined as the target traversal path, with anomaly point 9 as its endpoint. Next, local anomaly cluster region 3 (the third region) is processed by comparing the endpoint of the target traversal path of the preceding region (local anomaly cluster region 2) (anomaly point 9) with the two boundary points of local anomaly cluster region 3. It is confirmed that anomaly point 9 is connected to the starting boundary point (anomaly point 10). Therefore, the forward candidate traversal path of local anomaly cluster region 3 (anomaly point 10—anomaly point 11—anomaly point 12) is determined as the target traversal path. Finally, the target traversal paths for the remaining two regions are obtained, ensuring the continuity of the traversal direction between regions.

[0159] Step 30443: Connect the initial access path segment, the target traversal path of each local anomaly cluster area, and the inter-regional connection path segment of adjacent local anomaly cluster areas in sequence to obtain the target inspection path.

[0160] Optionally, the inspection device connects according to the inspection process sequence. In this embodiment of the invention, the initial access path segment is first connected, and the end point of the initial access path segment is connected to the starting point of the target traversal path of the initially reachable abnormal area. Then, according to the access order of abnormal areas, the end point of the target traversal path of the previous local abnormal cluster area is connected to the starting point of the target traversal path of the next local abnormal cluster area through the corresponding inter-regional connection path segment. Finally, the inspection device performs integrity verification on all connected paths to ensure that the paths are uninterrupted and without repetition, and can fully cover the initial access segment, all local abnormal cluster areas and inter-regional connection segments. After the verification is passed, the complete path is determined as the target inspection path. Therefore, the target inspection path can guide the inspection device to start from the current position and smoothly traverse all local abnormal cluster areas in sequence.

[0161] In one embodiment, the path segments to be connected include: initial access path segment (first current inspection position - anomaly point 1), target traversal path of local anomaly cluster area 1 (anomaly point 1 - anomaly point 2 - anomaly point 4), inter-regional connection path segment 1 (anomaly point 4 - anomaly point 7), target traversal path of local anomaly cluster area 2 (anomaly point 7 - anomaly point 8 - anomaly point 9), inter-regional connection path segment 2 (anomaly point 9 - anomaly point 10), and target traversal path of local anomaly cluster area 3 (anomaly point 10 - anomaly point 11 - anomaly point 12). The inspection device is connected sequentially according to the connection logic: the endpoint (anomaly point 1) of the initial access path segment (first current inspection position - anomaly point 1) is connected to the starting point (anomaly point 1) of the target traversal path of local anomaly cluster area 1; then, the endpoint (anomaly point 4) of the target traversal path of local anomaly cluster area 1 is connected to the starting point (anomaly point 7) of the target traversal path of local anomaly cluster area 2 through inter-regional connection path segment 1 (anomaly point 4 - anomaly point 7); then, the endpoint (anomaly point 9) of the target traversal path of local anomaly cluster area 2 is connected to the starting point (anomaly point 10) of the target traversal path of local anomaly cluster area 3 through inter-regional connection path segment 2 (anomaly point 9 - anomaly point 10). After the connection is completed, an integrity check is performed to confirm that the path is uninterrupted, without repetition, and fully covers all key sections, finally obtaining the target inspection path: first current inspection position - anomaly point 1 - anomaly point 2 - anomaly point 4 - anomaly point 7 - anomaly point 8 - anomaly point 9 - anomaly point 10 - anomaly point 11 - anomaly point 12.

[0162] The target inspection path constructed in this embodiment of the invention not only conforms to the physical layout of the line to be inspected, ensuring the smooth passage of the inspection device, but also achieves the optimization of path length and the maximization of inspection efficiency through precise matching of traversal directions. This solves the problems of direction confusion and path redundancy, improves the work efficiency of inspection in abnormal areas of the line, and thus ensures the dynamic and efficient operation of the inspection.

[0163] Optionally, the processes of steps 401 to 404 include:

[0164] Step 401: For each newly added anomaly, determine whether it has a physical connection relationship or spatial proximity relationship with any line anomaly point in any local anomaly cluster area in the anomaly point distribution relationship.

[0165] Optionally, for each newly added anomaly, a traversal judgment logic is executed: traverse each local anomaly cluster in the anomaly distribution relationship, extract the location information of all line anomalies within that cluster, and determine whether there is a physical connection or spatial proximity relationship between the newly added anomaly and each line anomaly within that cluster. Finally, record the judgment result for each newly added anomaly, i.e., whether it has at least one of the above two relationships with any line anomaly in any local anomaly cluster.

[0166] In one embodiment, the initial distribution of abnormal points includes three local abnormal clusters: local abnormal cluster 1 (containing abnormal points 1, 2, and 4), local abnormal cluster 2 (containing abnormal points 7, 8, and 9), and local abnormal cluster 3 (containing abnormal points 10, 11, and 12). The preset proximity threshold is 10 meters. When the inspection device inspects along the target inspection path, two new abnormal points are detected: new abnormal point 1 (located 3 meters away from abnormal point 4) and new abnormal point 2 (located in a branch section far from all existing local abnormal clusters, 15 meters away from the nearest abnormal point 12). The physical topology of the line shows that the line section where new abnormal point 1 is located is the same continuous line as the line section where abnormal point 4 is located, and the line section where new abnormal point 2 is located has no connection relationship with the line sections of all existing local abnormal clusters.

[0167] Furthermore, the inspection device makes separate judgments on the two newly added anomalies: For newly added anomaly 1, after traversing the three existing local anomaly clusters, it is found that it has a physical connection with anomaly 4 in local anomaly cluster 1, and the spatial straight-line distance of 3 meters is less than the proximity distance threshold of 10 meters (there is a spatial proximity relationship), so it is judged as "there is a connection"; For newly added anomaly 2, after traversing all local anomaly clusters, it is found that it has no physical connection with any line anomaly point in any cluster, and the spatial straight-line distance of 15 meters to the nearest anomaly 12 is greater than the proximity distance threshold of 10 meters (there is no spatial proximity relationship), so it is judged as "there is no connection".

[0168] Step 402: If an anomaly exists, add the new anomaly point to the corresponding first target local anomaly cluster area, and update the boundary points of the anomaly region of the first target local anomaly cluster area based on the new anomaly point to obtain the updated local anomaly cluster area. If an anomaly point does not exist, determine the new local anomaly candidate area.

[0169] Optionally, if the newly added anomaly is associated with any local anomaly cluster, then the local anomaly cluster is determined as the first target local anomaly cluster (i.e., the existing local anomaly cluster associated with the newly added anomaly), and the location information and related attributes of the newly added anomaly are added to the anomaly set of the cluster. At the same time, the boundary point of the anomaly area of ​​the cluster is adjusted based on the location of the newly added anomaly. If the newly added anomaly is located outside the edge of the original cluster and extends along the physical direction of the line, then the newly added anomaly is updated as the new boundary point of the anomaly area, resulting in the updated local anomaly cluster.

[0170] If a newly added anomaly is not associated with any existing local anomaly clusters, then the newly added anomaly is identified as a separate candidate region for new local anomalies.

[0171] In one embodiment, newly added anomaly point 1 is determined to be associated with local anomaly cluster area 1, and newly added anomaly point 2 is determined to be unrelated; the original boundary points of the local anomaly cluster area 1 are the starting point anomaly point 1 and the ending point anomaly point 4, the physical direction of the line is from anomaly point 1 to anomaly point 4, and newly added anomaly point 1 is located 3 meters outside anomaly point 4 along the line direction.

[0172] For the newly added anomaly point 1, the local anomaly cluster area 1 is determined as the first target local anomaly cluster area, and the newly added anomaly point 1 is added to the anomaly point set of this cluster area (the set becomes anomaly point 1, anomaly point 2, anomaly point 4, and newly added anomaly point 1). Since the newly added anomaly point 1 is located outside the original termination boundary point (anomaly point 4) and extends along the line direction, the original termination boundary point is updated to the newly added anomaly point 1, resulting in the updated local anomaly cluster area 1 (boundary point: anomaly point 1, newly added anomaly point 1). For the newly added anomaly point 2, since it has no correlation, it is determined as a new local anomaly candidate area, with the range centered on the newly added anomaly point 2, covering the line section within a 5-meter radius.

[0173] Step 403: Based on the target inspection path, perform boundary point coverage detection on the updated local anomaly cluster area and the newly added local anomaly candidate area. Local anomaly cluster areas or newly added local anomaly candidate areas whose boundary points of any anomaly area are not covered by the target inspection path are identified as uncovered anomaly areas.

[0174] Optionally, for each updated local anomaly cluster and each newly added local anomaly candidate area, boundary point coverage detection is performed: extract all anomaly area boundary points of the area, determine one by one whether the line segment where each boundary point is located is covered by the original target inspection path, and whether the distance between the path and the boundary point meets the coverage threshold requirement; finally, if any anomaly area boundary point of a certain area is not covered by the original target inspection path (i.e. does not meet the coverage judgment criteria), then the area is determined as an uncovered anomaly area. Therefore, an uncovered anomaly area refers to the area in the updated anomaly area or newly added candidate area that is not covered by the original target inspection path, and needs to be covered by path update.

[0175] In one embodiment, the new boundary point of the updated local anomaly cluster area 1 is anomaly point 1 and newly added anomaly point 1; the boundary point of the newly added local anomaly candidate area is newly added anomaly point 2; the line segment covered by the original target inspection path is the first current inspection position - anomaly point 1 - anomaly point 2 - anomaly point 4 - anomaly point 7 - anomaly point 8 - anomaly point 9 - anomaly point 10 - anomaly point 11 - anomaly point 12, and the preset coverage threshold is 2 meters.

[0176] For the updated local anomaly cluster 1, the original path covers the original termination boundary point (anomaly point 4), but the line segment containing the newly added termination boundary point (new anomaly point 1) is not included in the original path, and the closest distance to the original path is 3 meters (greater than the coverage threshold of 2 meters). Therefore, the updated local anomaly cluster 1 is determined to be an uncovered anomaly area. For the newly added local anomaly candidate area, the line segment containing its boundary point (new anomaly point 2) is not included in the original path, and the closest distance to the original path is 15 meters (greater than the coverage threshold of 2 meters). Therefore, the newly added local anomaly candidate area is determined to be an uncovered anomaly area. Finally, two uncovered anomaly areas were selected: the updated local anomaly cluster 1 and the newly added local anomaly candidate area.

[0177] Step 404: Based on the second current inspection position of the inspection device on the target inspection path and the uncovered abnormal area, update the target inspection path to obtain the updated inspection path.

[0178] Optionally, the target inspection path is updated based on the second current inspection position of the inspection device on the target inspection path and the uncovered abnormal area, to obtain the updated inspection path, as described in steps 4041 to 4044.

[0179] This invention, based on path updates between the current location and uncovered areas, ensures that the inspection device can start from the current location and cover newly added anomaly-related areas with the optimal path, while connecting unfinished sections of the original path. It balances path continuity and efficiency, solves the problem that fixed inspection paths cannot cope with newly added anomalies during the inspection process, realizes dynamic optimization of the inspection path, ensures the comprehensiveness of anomaly inspection, significantly improves the flexibility and reliability of line inspection work, and thus ensures dynamic and efficient inspection operations.

[0180] Optionally, the processes of steps 4041 to 4044 include:

[0181] Step 4041: Starting from the second current inspection position, traverse the transmission line topology based on the physical connection relationship of the line, and determine the first local anomaly cluster area or newly added local anomaly candidate area that is reached through the physical connection relationship of the continuous line in the uncovered anomaly area as the second target local anomaly cluster area.

[0182] Optionally, starting from the second current inspection position, based on the physical connection relationship indicated by the physical topology of the line, traverse all line segments corresponding to the uncovered abnormal areas, and determine whether each uncovered abnormal area can be reached through continuous physical connection relationships (i.e., in the path from the second current inspection position to the target area, all line segments are physically connected, without interruption or obstacles that prevent passage); finally, the first uncovered abnormal area that can be reached through continuous physical connection relationships (which may be the updated local abnormal cluster area or the newly added local abnormal candidate area) is determined as the second target local abnormal cluster area.

[0183] In one embodiment, the uncovered abnormal areas determined in step 403 are two: the updated local abnormal cluster area 1 (boundary point: abnormal point 1, newly added abnormal point 1) and the newly added local abnormal candidate area (boundary point: newly added abnormal point 2); the second current inspection position of the inspection device is the position of abnormal point 2 on the original target inspection path; the physical topology of the line shows that, starting from abnormal point 2, the updated local abnormal cluster area 1 (abnormal point 2 belongs to the original abnormal point set of this cluster area) can be directly reached through a continuous physical line connection, while reaching the newly added local abnormal candidate area requires passing through a section of line interruption without physical connection, and cannot be directly reached through a continuous physical line connection.

[0184] Starting from the second current inspection position (anomaly point 2), the accessibility of the two uncovered anomaly areas is traversed: the updated local anomaly cluster area 1 can be directly reached through a continuous physical connection, and it is the first reachable area; the newly added local anomaly candidate area cannot be directly reached through a continuous physical connection. Therefore, the updated local anomaly cluster area 1 is determined as the second target local anomaly cluster area.

[0185] Step 4042: Based on the physical connection relationship between the second current inspection position and the boundary point of the abnormal area at the starting end of the second target local abnormal cluster area, the continuous line segment from the second current inspection position to the boundary point of the abnormal area at the starting end is determined as the insertion access path segment.

[0186] Optionally, the inspection device queries the physical topology of the line to confirm the physical connection between the second current inspection position and the boundary point of the starting abnormal area of ​​the second target local abnormal cluster area, and filters out the continuous line section between the two that can ensure the smooth passage of the inspection device (i.e., the line and auxiliary passage section without line interruption, without obstacles, and meeting the passage requirements of the inspection device).

[0187] Furthermore, the inspection device identifies this continuous line segment as the insertion access path segment. Therefore, the insertion access path segment represents the connecting path segment that connects the second current inspection position of the inspection device with the second target local anomaly cluster area.

[0188] In one embodiment, the second target local anomaly cluster area determined in step 4041 is the updated local anomaly cluster area 1, and its starting point of the anomaly area boundary point is anomaly point 1; the second current inspection position of the inspection device is anomaly point 2; the physical topology of the line shows that there is a continuous line segment between anomaly point 2 and anomaly point 1 (i.e., the segment from anomaly point 1 to anomaly point 2 in the original target inspection path), and this segment is unobstructed and meets the passage requirements of the inspection device. After querying and confirming the above connection relationship and continuous line segment, the continuous line segment from the second current inspection position (anomaly point 2) to the starting point of the second target local anomaly cluster area boundary point (anomaly point 1) is determined as the insertion access path segment.

[0189] Step 4043: If the second target local anomaly cluster area is the updated local anomaly cluster area, then a supplementary traversal path is generated based on the anomaly point sequence path corresponding to the updated local anomaly cluster area. If the second target local anomaly cluster area is a newly added local anomaly candidate area and contains only a single newly added anomaly point, then the line segment where the newly added anomaly point is located is used as the supplementary traversal path. If the second target local anomaly cluster area is a newly added local anomaly candidate area and contains multiple newly added anomaly points, then its anomaly point sequence path is generated according to the line physical connection relationship, and the anomaly point sequence path is used as the supplementary traversal path.

[0190] Optionally, if the second target local anomaly cluster area is the updated local anomaly cluster area, then extract its updated anomaly point sequence path and determine the path as the supplementary traversal path.

[0191] Optionally, if the second target local anomaly cluster area is a newly added local anomaly candidate area and contains only a single newly added anomaly point, then the line segment where the newly added anomaly point is located, that is, the line segment centered on the newly added anomaly point and covering the surrounding preset range, is used as the supplementary traversal path. The preset range is determined according to the inspection accuracy requirements.

[0192] Optionally, if the second target local anomaly cluster area is a newly added local anomaly candidate area and contains multiple newly added anomaly points, then based on the physical connection relationship of the line indicated by the physical topology data of the line, the multiple newly added anomaly points are sorted according to the physical direction of the line to generate an anomaly point sequence path, and the anomaly point sequence path is determined as the supplementary traversal path.

[0193] In one embodiment, the second target local anomaly cluster area is the updated local anomaly cluster area 1, and its updated anomaly point sequence path is anomaly point 1—anomaly point 2—anomaly point 4—new anomaly point 1 (sorted according to the physical route of the line). If the second target local anomaly cluster area is a newly added local anomaly candidate area, and this candidate area contains only a single newly added anomaly point 2, then the line segment covering a 5-meter radius around the newly added anomaly point 2 is used as a supplementary traversal path. If the candidate area contains two anomaly points, newly added anomaly point 3 and newly added anomaly point 4, and the physical route of the line is from newly added anomaly point 3 to newly added anomaly point 4, then an anomaly point sequence path (new anomaly point 3—new anomaly point 4) is generated according to this route and used as a supplementary traversal path. For the updated local anomaly cluster area 1, its updated anomaly point sequence path (anomaly point 1—anomaly point 2—anomaly point 4—new anomaly point 1) is extracted, and this path is determined as the supplementary traversal path.

[0194] Step 4044: Based on the inserted access path segment, the supplementary traversal path, and the unexecuted path after the second current inspection position in the target inspection path, the updated inspection path is obtained by concatenating the paths.

[0195] Optionally, the paths are connected sequentially according to the inspection execution order. First, the starting point of the inserted access path segment is connected to the second current inspection position, and the ending point of the inserted access path segment is connected to the starting point of the supplementary traversal path (the starting boundary point of the second target local anomaly cluster area). Then, the ending point of the supplementary traversal path is connected to the starting point of the unexecuted path after the second current inspection position in the original target inspection path. Finally, the complete path after splicing is checked for integrity and continuity to ensure that the path is uninterrupted and without repetition, and can fully cover the inserted access segment, the second target local anomaly cluster area, and the remaining unexecuted segment of the original path. After the check passes, the complete path is determined as the updated inspection path.

[0196] In one embodiment, the inserted access path segment is anomaly point 2—anomaly point 1; the supplementary traversal path obtained in step 4043 is anomaly point 1—anomaly point 2—anomaly point 4—new anomaly point 1; the path that has not been executed after the second current inspection position (anomaly point 2) in the original target inspection path is anomaly point 2—anomaly point 4—anomaly point 7—anomaly point 8—anomaly point 9—anomaly point 10—anomaly point 11—anomaly point 12.

[0197] Connect the points sequentially according to the splicing logic: connect the starting point of the inserted access path segment (anomaly point 2) with the second current inspection position, and connect the ending point of the inserted access path segment (anomaly point 1) with the starting point of the supplementary traversal path (anomaly point 1); then connect the ending point of the supplementary traversal path (new anomaly point 1) with the starting point of the unexecuted path of the original target inspection path (anomaly point 2) (wherein, the supplementary traversal path has covered the section from anomaly point 2 to anomaly point 4, and after connection, the repeated section from anomaly point 2 to anomaly point 4 in the original path can be skipped); the final updated inspection path is: anomaly point 2—anomaly point 1—anomaly point 2—anomaly point 4—new anomaly point 1—anomaly point 7—anomaly point 8—anomaly point 9—anomaly point 10—anomaly point 11—anomaly point 12.

[0198] This invention integrates the supplementary path with the remaining segments of the original path, avoiding path duplication and interruption, ensuring the continuity of inspection tasks, solving the problem of being unable to cope with new anomalies, realizing dynamic adaptive adjustment of the inspection path, ensuring that all uncovered anomaly areas can be accurately covered, while maximizing the use of the original inspection path, reducing path redundancy, and improving the efficiency of inspection work.

[0199] Furthermore, the intelligent inspection device based on machine vision provided by the present invention will be described below. The intelligent inspection device based on machine vision described below and the intelligent inspection method based on machine vision described above can be referred to in correspondence.

[0200] Optionally, refer to Figure 2 , Figure 2 This is a schematic diagram of the intelligent inspection device based on machine vision provided by the present invention. The intelligent inspection device based on machine vision includes:

[0201] The anomaly identification module 210 is used to acquire images of the visible section of the line to be inspected based on the machine vision device, obtain full-area image data, and identify line anomalies based on the full-area image data to obtain anomaly location information.

[0202] The distribution relationship construction module 220 is used to construct the distribution relationship of abnormal points based on the spatial proximity relationship and physical connection relationship between various abnormal points of the line obtained by analyzing the location information of abnormal points;

[0203] The inspection path decision module 230 is used to make inspection path decisions based on the first current inspection position of the inspection device and the distribution relationship of abnormal points, and to determine the target inspection path.

[0204] The inspection path update module 240 is used to update the target inspection path based on the newly added abnormal points and the distribution relationship of the abnormal points if the inspection device detects that there are new abnormal points during the inspection process along the target inspection path, so as to obtain the updated inspection path and carry out inspection along the updated inspection path.

[0205] The embodiments of the present invention solve the problems of easily overlooking dynamically occurring anomalies, delaying processing opportunities, and increasing inspection time. They enable the inspection path to dynamically adapt to the actual line status during the inspection process, realize the rational allocation of inspection resources, improve inspection efficiency, and dynamically cover abnormal points, thus ensuring the dynamic and efficient operation of inspection.

[0206] Please see Figure 3 , Figure 3 An embodiment diagram of an electronic device provided in accordance with the present invention. For example... Figure 3 As shown, this embodiment of the invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, it performs the following steps:

[0207] The machine vision equipment is used to acquire images of the visible section of the line to be inspected, obtain full-area image data, and identify line anomalies based on the full-area image data to obtain the location information of the anomalies.

[0208] Based on the spatial proximity relationship and physical connection relationship between anomalies on each line obtained from the analysis of anomaly location information, anomaly distribution relationship is constructed.

[0209] Based on the relationship between the first current inspection position of the inspection device and the distribution of abnormal points, the inspection path is determined to identify the target inspection path.

[0210] If a new anomaly is detected during the inspection process along the target inspection path, the target inspection path is updated based on the new anomaly and its distribution relationship, and the inspection is carried out along the updated inspection path.

[0211] Please see Figure 4 , Figure 4 An embodiment diagram of a computer-readable storage medium provided in accordance with an embodiment of the present invention is shown. Figure 4 As shown, this embodiment provides a computer-readable storage medium 400 on which a computer program 311 is stored. When the computer program 311 is executed by a processor, it performs the following steps:

[0212] The machine vision equipment is used to acquire images of the visible section of the line to be inspected, obtain full-area image data, and identify line anomalies based on the full-area image data to obtain the location information of the anomalies.

[0213] Based on the spatial proximity relationship and physical connection relationship between anomalies on each line obtained from the analysis of anomaly location information, anomaly distribution relationship is constructed.

[0214] Based on the relationship between the first current inspection position of the inspection device and the distribution of abnormal points, the inspection path is determined to identify the target inspection path.

[0215] If a new anomaly is detected during the inspection process along the target inspection path, the target inspection path is updated based on the new anomaly and its distribution relationship, and the inspection is carried out along the updated inspection path.

[0216] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the intelligent inspection method based on machine vision provided by the above methods, the method including:

[0217] The machine vision equipment is used to acquire images of the visible section of the line to be inspected, obtain full-area image data, and identify line anomalies based on the full-area image data to obtain the location information of the anomalies.

[0218] Based on the spatial proximity relationship and physical connection relationship between anomalies on each line obtained from the analysis of anomaly location information, anomaly distribution relationship is constructed.

[0219] Based on the relationship between the first current inspection position of the inspection device and the distribution of abnormal points, the inspection path is determined to identify the target inspection path.

[0220] If a new anomaly is detected during the inspection process along the target inspection path, the target inspection path is updated based on the new anomaly and its distribution relationship, and the inspection is carried out along the updated inspection path.

[0221] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0222] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0223] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A machine vision-based intelligent inspection method, characterized in that, include: The machine vision equipment is used to acquire images of the visible section of the line to be inspected, and full-area image data is obtained. Based on the full-area image data, line anomaly points are identified to obtain anomaly point location information. Based on the spatial proximity relationship and physical connection relationship between each line anomaly point obtained from the analysis of the anomaly point location information, the distribution relationship of the anomaly points is constructed. Based on the first current inspection position of the inspection device and the distribution relationship of the abnormal points, the inspection path is determined to identify the target inspection path. If the inspection device detects new abnormal points during the inspection process along the target inspection path, the target inspection path is updated based on the new abnormal points and the distribution relationship of the abnormal points to obtain the updated inspection path, and the inspection is carried out along the updated inspection path. The steps for obtaining the updated inspection path include: For each newly added anomaly, determine whether it has a physical connection or spatial proximity with any line anomaly in any local anomaly cluster area in the anomaly distribution relationship. If it exists, the new anomaly point is added to the corresponding first target local anomaly cluster area, and the boundary points of the anomaly region of the first target local anomaly cluster area are updated based on the new anomaly point to obtain the updated local anomaly cluster area; if it does not exist, the new anomaly point is determined as a new local anomaly candidate area. Based on the target inspection path, boundary point coverage detection is performed on the updated local anomaly cluster area and the newly added local anomaly candidate area. Any local anomaly cluster area or newly added local anomaly candidate area whose boundary point is not covered by the target inspection path is determined as an uncovered anomaly area. The updated inspection path is obtained by updating the target inspection path based on the second current inspection position of the inspection device on the target inspection path and the uncovered abnormal area. The steps involved in determining the target inspection path include: Based on the distribution relationship of the abnormal points, the local abnormal cluster area that is directly connected to the first current inspection position through the physical connection of the line by the boundary point of the abnormal area is determined as the initially reachable abnormal area. Starting from the initial reachable abnormal region, based on the physical connection relationship of the line indicated by the abnormal distribution topology in the abnormal point distribution relationship, the remaining local abnormal clusters directly connected to it are traversed in sequence to obtain the abnormal region adjacency sequence. The local abnormal clusters in the abnormal region adjacency sequence are sorted according to their physical arrangement on the transmission line to obtain the abnormal region access order that extends unidirectionally along the physical connection relationship of the line from the initial reachable abnormal region. Based on the boundary points of the abnormal regions of two adjacent local abnormal clusters in the abnormal region access sequence, the inspection path is determined to identify the target inspection path.

2. The intelligent inspection method based on machine vision according to claim 1, characterized in that, The process of updating the target inspection path based on the second current inspection position of the inspection device on the target inspection path and the uncovered abnormal area to obtain the updated inspection path includes: Starting from the second current inspection position, based on the physical connection relationship of the line, the transmission line topology is traversed, and the first local anomaly cluster area or newly added local anomaly candidate area reached through the physical connection relationship of the continuous line in the uncovered anomaly area is determined as the second target local anomaly cluster area. Based on the physical connection relationship between the second current inspection position and the boundary point of the abnormal area at the starting end of the second target local abnormal cluster area, the continuous line segment from the second current inspection position to the boundary point of the abnormal area at the starting end is determined as the insertion access path segment; If the second target local anomaly cluster area is the updated local anomaly cluster area, then a supplementary traversal path is generated based on the anomaly point sequence path corresponding to the updated local anomaly cluster area; if the second target local anomaly cluster area is a newly added local anomaly candidate area and contains only a single newly added anomaly point, then the line segment where the newly added anomaly point is located is used as the supplementary traversal path; if the second target local anomaly cluster area is a newly added local anomaly candidate area and contains multiple newly added anomaly points, then its anomaly point sequence path is generated according to the line physical connection relationship, and the anomaly point sequence path is used as the supplementary traversal path. The updated inspection path is obtained by concatenating the inserted access path segment, the supplementary traversal path, and the unexecuted path after the second current inspection position in the target inspection path.

3. The intelligent inspection method based on machine vision according to claim 1, characterized in that, The step of determining the target inspection path by making inspection path decisions based on the boundary points of two adjacent local anomaly clusters in the anomaly region access sequence includes: For two adjacent local abnormal clusters in the abnormal area access sequence, the connection path segment between the regions is determined based on the physical connection relationship between the boundary point of the abnormal area corresponding to the end of the previous local abnormal cluster and the boundary point of the abnormal area corresponding to the beginning of the next local abnormal cluster. Based on the physical connection relationship between the first current inspection location and the boundary point of the abnormal area corresponding to the starting end of the initially reachable abnormal area, a continuous line segment from the first current inspection location to the boundary point of the abnormal area corresponding to the starting end of the initially reachable abnormal area is determined to obtain the initial access path segment. For each local anomaly cluster in the abnormal region access sequence, bidirectional candidate traversal paths with opposite traversal directions are generated based on its corresponding anomaly point sequence path; the first candidate path traverses from the anomaly region boundary point at the beginning to the anomaly region boundary point at the end, and the second candidate path traverses from the anomaly region boundary point at the end to the anomaly region boundary point at the beginning. The inspection path is determined by making inspection path decisions based on the initial access path segment, the inter-regional connection path segments of adjacent local anomaly clusters, and the bidirectional candidate traversal path of each local anomaly cluster.

4. The intelligent inspection method based on machine vision according to claim 3, characterized in that, Based on the initial access path segment, the inter-regional connection path segments of adjacent local anomaly clusters, and the bidirectional candidate traversal path of each local anomaly cluster, inspection path decision is made to determine the target inspection path, including: The endpoint of the initial access path segment is matched with the two boundary points of the abnormal region of the initial reachable abnormal region. If the endpoint of the initial access path segment is connected to the boundary point of the abnormal region at the beginning of the initial reachable abnormal region through a physical connection, the candidate traversal path in the traversal direction from the beginning to the end is determined as the target traversal path. If the endpoint of the initial access path segment is connected to the boundary point of the abnormal region at the end of the initial reachable abnormal region, the candidate traversal path in the traversal direction from the end to the beginning is determined as the target traversal path. For each local anomaly cluster starting from the second local anomaly cluster in the anomaly region access sequence, the endpoint of the candidate traversal path of the previous local anomaly cluster is compared with the two anomaly region boundary points of the current local anomaly cluster. If the endpoint is connected to the anomaly region boundary point of the starting end of the current local anomaly cluster through a physical connection, the candidate traversal path in the traversal direction from the starting end to the ending end is determined as the target traversal path. If the endpoint is connected to the anomaly region boundary point of the ending end, the candidate traversal path in the traversal direction from the ending end to the starting end is determined as the target traversal path. In this way, the endpoints of the candidate traversal paths are compared with the two anomaly region boundary points to obtain the target traversal paths for the remaining local anomaly clusters. The target inspection path is obtained by sequentially connecting the initial access path segment, the target traversal path of each local anomaly cluster area, and the inter-regional connection path segment of adjacent local anomaly cluster areas.

5. The intelligent inspection method based on machine vision according to any one of claims 1 to 4, characterized in that, The spatial proximity and physical connection relationships between various line anomalies obtained from the analysis of the anomaly location information are used to construct the anomaly distribution relationship, including: For any two different first line anomaly points and second line anomaly points, if there is at least one relationship between the first line anomaly point and the second line anomaly point, either a spatial proximity relationship or a physical connection relationship, then the first line anomaly point and the second line anomaly point are determined as an anomaly point pair. Merge pairs of anomalies that share the same line anomaly point to obtain adjacent anomaly point groups; any two line anomalies within an adjacent anomaly point group can be connected to each other through the path in the anomaly point pair. For each adjacent anomaly point group, an anomaly point subgraph is constructed with each line anomaly point in the group as a node and the physical connection relationship or spatial proximity relationship between each line anomaly point in the group as an edge. The distribution relationship of the abnormal points is constructed based on the route indicated by the physical connection relationship of the route in the connected subgraph of each abnormal point.

6. The intelligent inspection method based on machine vision according to claim 5, characterized in that, The method of constructing the distribution relationship of abnormal points based on the line routes indicated by the physical connection relationships in the connectivity subgraph of each abnormal point includes: Based on the route indicated by the physical connection relationship of the line in the connected subgraph of each anomaly point, adjacent line anomaly points are connected in sequence to obtain the anomaly point sequence path; there is a physical connection relationship between adjacent line anomaly points in each anomaly point sequence path. For any target path in the anomaly point sequence path, if the number of line anomalies it contains is greater than or equal to a preset minimum anomaly point threshold, then the line segment covered by the target path is determined as a local anomaly cluster area. For each local anomaly cluster, the line anomaly points corresponding to its starting and ending ends are taken as the boundary points of the anomaly region. Combining the connection relationship between each local anomaly cluster in the anomaly point connectivity subgraph, an anomaly distribution topology is constructed. The anomaly distribution topology represents the interconnection relationship between each local anomaly cluster on the line to be inspected. The abnormal point distribution relationship is obtained by integrating the local abnormal clusters, abnormal region boundary points and their interconnections contained in the abnormal distribution topology.

7. An intelligent inspection device based on machine vision, characterized in that, The apparatus is used to implement the machine vision-based intelligent inspection method as described in any one of claims 1 to 6; the apparatus includes: The anomaly identification module is used to acquire images of the visible section of the line to be inspected based on machine vision equipment, obtain full-area image data, and identify line anomalies based on the full-area image data to obtain anomaly location information. The distribution relationship construction module is used to construct the distribution relationship of abnormal points based on the spatial proximity relationship and physical connection relationship between each abnormal point of the line obtained by analyzing the abnormal point location information. The inspection path decision module is used to make inspection path decisions based on the first current inspection position of the inspection device and the distribution relationship of the abnormal points, and to determine the target inspection path. The inspection path update module is used to update the target inspection path based on the new abnormal point and the distribution relationship of the abnormal point if the inspection device detects a new abnormal point during the inspection along the target inspection path, so as to obtain the updated inspection path and carry out the inspection along the updated inspection path. The steps for obtaining the updated inspection path include: For each newly added anomaly, determine whether it has a physical connection or spatial proximity with any line anomaly in any local anomaly cluster area in the anomaly distribution relationship. If it exists, the new anomaly point is added to the corresponding first target local anomaly cluster area, and the boundary points of the anomaly region of the first target local anomaly cluster area are updated based on the new anomaly point to obtain the updated local anomaly cluster area; if it does not exist, the new anomaly point is determined as a new local anomaly candidate area. Based on the target inspection path, boundary point coverage detection is performed on the updated local anomaly cluster area and the newly added local anomaly candidate area. Any local anomaly cluster area or newly added local anomaly candidate area whose boundary point is not covered by the target inspection path is determined as an uncovered anomaly area. The updated inspection path is obtained by updating the target inspection path based on the second current inspection position of the inspection device on the target inspection path and the uncovered abnormal area. The steps involved in determining the target inspection path include: Based on the distribution relationship of the abnormal points, the local abnormal cluster area that is directly connected to the first current inspection position through the physical connection of the line by the boundary point of the abnormal area is determined as the initially reachable abnormal area. Starting from the initial reachable abnormal region, based on the physical connection relationship of the line indicated by the abnormal distribution topology in the abnormal point distribution relationship, the remaining local abnormal clusters directly connected to it are traversed in sequence to obtain the abnormal region adjacency sequence. The local abnormal clusters in the abnormal region adjacency sequence are sorted according to their physical arrangement on the transmission line to obtain the abnormal region access order that extends unidirectionally along the physical connection relationship of the line from the initial reachable abnormal region. Based on the boundary points of the abnormal regions of two adjacent local abnormal clusters in the abnormal region access sequence, the inspection path is determined to identify the target inspection path.

8. An electronic device, comprising: Memory, used to store computer software programs; A processor for reading and executing the computer software program, characterized in that, when the processor executes the computer software program, it implements the intelligent inspection method based on machine vision as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium, wherein a computer software program is stored therein, characterized in that, When the computer software program is executed by the processor, it implements the intelligent inspection method based on machine vision as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes a computer program, which, when executed by a processor, implements the machine vision-based intelligent inspection method as described in any one of claims 1 to 6.

Images