Photovoltaic station maintenance path planning method and device and computer readable storage medium

Abstract

The invention belongs to the technical field of photovoltaic power generation inspection, and particularly relates to a photovoltaic station maintenance path planning method and device and a computer readable storage medium. The method comprises the following steps: acquiring an orthophoto map of a photovoltaic station through an unmanned aerial vehicle, constructing a photovoltaic string detection model, and accurately identifying the position of a photovoltaic panel string; constructing a walkable maintenance area around the detected photovoltaic string, extracting a path skeleton and determining a node connection relationship; planning a maintenance path by using a shortest path algorithm based on the position of the fault point, and optimizing the length of the path through a node stretching algorithm; finally, the method is deployed to terminal equipment, and the maintenance path of the photovoltaic station is autonomously planned. According to the method, the problems that the traditional maintenance path planning efficiency is low, the maintenance difficulty of a fault photovoltaic string in a large-scale photovoltaic field station is relatively high, and the operation and maintenance efficiency of the photovoltaic field station is relatively low are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic power generation inspection, and more specifically, relates to a method and device for planning a maintenance path of a photovoltaic power station and a computer-readable storage medium. Background Art

[0002] Photovoltaic power stations usually consist of a large number of photovoltaic panel strings, with a wide distribution range and a complex structure. During the daily maintenance and repair of photovoltaic power stations, maintenance personnel need to quickly and accurately plan the optimal repair path to improve the repair efficiency and reduce the operation and maintenance costs. The current repair path planning of photovoltaic power stations mainly relies on manual planning and uses the shortest path planning algorithm to complete the path planning for repairing faulty photovoltaic strings, but the problems are also obvious. First of all, the manual planning of the repair path completely depends on the construction experience of the maintenance personnel. When facing large-scale photovoltaic power stations and multi-point faulty photovoltaic panels, it is very difficult to achieve the global shortest path in the manually planned repair path, which affects the efficiency during the repair of photovoltaic power stations. In addition, using the shortest path planning algorithm is a relatively appropriate idea. Some solutions directly perform path planning at the pixel level on the original DOM image of the photovoltaic power station. The path planning efficiency of this method is low, and it takes a huge amount of time for the repair path planning of large-scale photovoltaic power stations, and it is even impossible to be applied to actual tasks. There are also some works that perform repair path planning based on the known positions of photovoltaic strings and combine the shortest path planning algorithm, but they cannot reasonably organize the path connection relationship between photovoltaic strings, making it difficult to carry out subsequent path planning. It can be seen that the processing of the original photovoltaic power station before path planning is particularly crucial.

[0003] Chinese invention patent CN114462665A discloses a method and system for planning a repair path of a photovoltaic power station. The invention obtains the parameter information of each of the N repair components; uses a preset algorithm to calculate the repair priority order of the N repair components in combination with the parameter information of each of the N repair components; and generates a repair path diagram corresponding to the N repair components according to the repair priority order of the N repair components. The present invention obtains the parameter information of each of the N repair components, then calculates the repair priority order of the N repair components by using a preset algorithm, and then generates a repair path diagram corresponding to the N repair components.

[0004] In summary, the prior art provides the position information of faulty photovoltaic strings, but does not plan the repair path. The repair of faulty photovoltaic strings in large-scale photovoltaic power stations is difficult, and the operation and maintenance efficiency of photovoltaic power stations is low. Summary of the Invention

[0005] The present invention aims to overcome at least one defect of the above-mentioned prior art, and provides a method for planning the maintenance path of a photovoltaic power station, so as to solve the problems of low efficiency, poor adaptability, and insufficient accuracy in traditional maintenance path planning, significantly improve the maintenance efficiency and operation and maintenance management level of the photovoltaic power station, and is applicable to photovoltaic power stations with complex terrains and equipment distributions.

[0006] In another aspect of the present invention, a device for the method of planning the maintenance path of a photovoltaic power station is provided, which executes a method for planning the maintenance path of a photovoltaic power station as described above.

[0007] And, a computer-readable storage medium executes a method for planning the maintenance path of a photovoltaic power station as described above.

[0008] The detailed technical solution of the present invention is as follows: A method for planning the maintenance path of a photovoltaic power station, the method comprising: S1. Collect the orthophoto image of the photovoltaic power station and synthesize the ground DOM image of the photovoltaic power station; S2. Detect the photovoltaic panel strings in the ground DOM image of the photovoltaic power station through the photovoltaic string detection model of the photovoltaic power station DOM image; S3. Construct a walkable maintenance area around the detected photovoltaic strings, and obtain the skeleton of the walkable path and the connection relationship between the skeleton nodes; S4. Determine the position where the fault point is located in the walkable maintenance area, and plan the shortest maintenance path of the photovoltaic power station based on the shortest path algorithm; S5. Determine the nodes passed through in the planned maintenance path and the pixel length between adjacent nodes, stretch and optimize the path between adjacent nodes, and deploy it to the terminal hardware device.

[0009] Further, the S2 specifically includes: S21. Collect the ground DOM image of the photovoltaic power station and construct a training data set; S22. Construct a photovoltaic string detection model by using a deep learning algorithm; S23. Label the photovoltaic strings in the training data set through a labeling tool; S24. Train the photovoltaic string detection model through the training data set and optimize the model parameters; S25. Input the ground DOM image into the trained photovoltaic string detection model, and the model detects and outputs the position and bounding box information of the photovoltaic strings; S26. Perform post-processing on the detection results, remove redundant detection frames with high overlap through the non-maximum suppression algorithm to reduce false detections; at the same time, set a confidence threshold to filter out detection results with low confidence; then, map the obtained bounding boxes to an image mask to obtain the photovoltaic string mask image in the photovoltaic power station.

[0010] Further, S3 specifically includes: S31. Perform a merging process on adjacent photovoltaic strings, fill the gaps between the photovoltaic strings, and expand the area around the photovoltaic strings as a walkable area; S32. Refine the walkable area to obtain a two-dimensional skeleton image of the walking path. The pixel width of the path skeleton is 1, and the connection points of each section of the skeleton are skeleton nodes; S33. Detect bifurcation nodes in the refined skeleton image and calculate the connectivity relationship between the nodes.

[0011] Further, the merging process includes: performing a morphological closing operation on adjacent strings with a spacing < 1.5 m, and the kernel size is 15×15 pixels; The expansion of the area around the photovoltaic strings as a walkable area includes: expanding 2 meters along the string boundary to generate a walkable area surrounding the photovoltaic strings.

[0012] Further, the refinement of the walkable area includes: using the Zhang-Suen parallel refinement algorithm, iterating 12 times to obtain the skeleton image, extracting 386 nodes and 452 edges, and constructing an adjacency matrix to store the connectivity relationship.

[0013] Further, S4 specifically includes: Taking the points on the maintenance path in front of the faulty photovoltaic string as maintenance points; When specifying the starting maintenance point, use local graph theory modeling, solve it using the nearest neighbor algorithm combined with a dynamic adjustment strategy, generate a local shortest maintenance path, and plan the local shortest maintenance path of the photovoltaic power station; When not specifying the starting maintenance point, use global graph theory modeling, solve it using the nearest neighbor algorithm, and at the same time use multi-objective optimization and path segmenting methods, combined with dynamic update and parallelization strategies, to generate a global shortest maintenance path and plan the global shortest maintenance path of the photovoltaic power station.

[0014] Further, determine the nodes passed through in the planned maintenance path and the pixel lengths between adjacent nodes, and stretch and optimize the paths between adjacent nodes. Specifically, it includes: S51. Set the starting point of the path as point P; S52. Traverse the pixel points C passed through by the maintenance path after point P one by one, and connect the pixel point C to point P with a straight line; S53. Check whether the connected straight line crosses the mask area of the photovoltaic string; S54. If it crosses, stop traversing, take the straight line between point P and point C as the optimized path after stretching between the two points, update the pixel point C to P, and continue to iterate steps S52 and S53; S55. If not crossing, continue to iterate through steps S52 and S53; S56. If pixel C is the end point of the path, iteratively execute steps S51 - S55 until the length of the optimized maintenance path does not change. The obtained optimized maintenance path is the optimal maintenance path; S57. Convert the DOM image pixels of each node in the optimal maintenance path into ground longitude and latitude coordinates according to the engineering coordinate system stored in the DOM image, providing navigation guidance for the fault string detection of the photovoltaic power station.

[0015] Further, the deployment to the terminal hardware device includes: First, export the photovoltaic string detection model as an onnx - format model, and use the optimization tool of ONNX to simplify the exported ONNX model; Second, use RKNN Toolkit to load the optimized ONNX model, perform quantization processing on the ONNX model, and convert the floating - point model into a fixed - point rknn model; Then, deploy the generated rknn model to the target hardware platform, and optimize the model or hardware configuration according to the actual running performance; Finally, convert the other parts of the method into a terminal program, deploy it to the RV1126 platform, and achieve autonomous maintenance path planning on the terminal platform.

[0016] Further, S1 also includes: pre - processing the collected ortho - image sequence of the photovoltaic power station, including denoising, image enhancement, and geometric correction.

[0017] In another aspect of the present invention, there is provided a device for a photovoltaic power station maintenance path planning method, and the device includes: At least one processor; and A memory that stores instructions, and when the instructions are executed by the at least one processor, the at least one processor executes a photovoltaic power station maintenance path planning method as described above.

[0018] In another aspect of the present invention, there is also provided a computer - readable storage medium that stores executable instructions, and when the instructions are executed, the machine executes a photovoltaic power station maintenance path planning method as described above.

[0019] Compared with the prior art, the beneficial effects of the present invention are: (1) A method, device and computer-readable medium for planning the maintenance path of a photovoltaic power station provided by the present invention, a method for planning the maintenance path of a photovoltaic power station that can be deployed at the terminal. Compared with the traditional maintenance path planning scheme, this scheme does not require manual parameters and can automatically complete the path planning for repairing faulty photovoltaic strings in a photovoltaic power station at low cost, greatly reducing the walking of ineffective paths by maintenance personnel and improving the maintenance operation efficiency of large-scale photovoltaic power stations; when the method is deployed on a terminal device, after a potential fault point appears, it can independently perform path planning without cloud data transmission and output a navigation scheme for the maintenance path planning result, improving the operation and maintenance efficiency of the photovoltaic power station and further liberating productivity.

[0020] (2) A method, device and computer-readable medium for planning the maintenance path of a photovoltaic power station provided by the present invention, a method for refining the skeleton of a feasible walking path in the method for planning the maintenance path of a photovoltaic power station, simplifies the actual walking path area into a one-dimensional skeleton representation, greatly reducing the huge cost of directly performing path planning based on image pixels and improving the efficiency of path planning.

[0021] (3) A method, device and computer-readable medium for planning the maintenance path of a photovoltaic power station provided by the present invention, a path planning method based on node stretching in the published shortest maintenance path planning, optimizes the path between nodes through stretching to adapt to the actual maintenance walking area of the power station, realizing the planning of the actual shortest feasible walking path; only a low-cost edge computing platform is required to realize autonomous maintenance path planning, with a large improvement in efficiency and high accuracy of path planning compared with the traditional maintenance path planning scheme. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flowchart of a method for planning the maintenance path of a photovoltaic power station according to the present invention.

[0023] Figure 2 It is an orthoimage DOM image in Embodiment 1 of the present invention.

[0024] Figure 3 It is a schematic diagram of obtaining a photovoltaic string mask image in a photovoltaic power station in Embodiment 1 of the present invention.

[0025] Figure 4 It is a schematic diagram of a feasible walking area in Embodiment 1 of the present invention.

[0026] Figure 5 It is a schematic diagram of a skeleton node in Embodiment 1 of the present invention.

[0027] Figure 6 It is a schematic diagram of the optimized maintenance path, which is the optimal maintenance path, in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0029] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0030] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0031] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0032] Embodiment 1 Refer Figure 1 , this embodiment provides a method for planning the maintenance path of a photovoltaic power station. A 10MW photovoltaic power station is located in a certain area in the central part of Shandong, covering an area of about 30 hectares and including 3,200 groups of photovoltaic panel strings. There are 5 fault points in this power station recently, with coordinates A, B, C, D, and E respectively, and a rapid repair path needs to be planned. The specific implementation process of the method includes: S1. The UAV collects the orthophoto image of the photovoltaic power station and synthesizes the ground DOM image of the photovoltaic power station by using the orthophoto image; The UAV is used to carry out aerial photography of the photovoltaic power station with a high-resolution camera. Through the UAV flight control system, the attitude of the camera is ensured to be stable during the aerial photography process to avoid image distortion. In addition, the orthophoto image sequence of the photovoltaic power station collected by the UAV is preprocessed, including denoising, image enhancement, and geometric correction, to ensure that the image quality meets the requirements of subsequent analysis. The DOM synthesis software is used to synthesize the orthophoto DOM image of the photovoltaic power station, as Figure 2 shown; Preferably, the DJI M300 RTK UAV is used to carry the Hasselblad H6D camera to collect the orthophoto image of the photovoltaic power station, with a flight height of 80 meters and an overlap rate of 80%, and the orthophoto image sequence of the power station is collected; then, the geometric correction and splicing are carried out by the DJI Terra software to generate a DOM image with a resolution of 2cm / pixel, and the coordinate system is WGS84-UTM Zone 48N.

[0033] S2. Construct a training data set and a photovoltaic string detection model and train them: The photovoltaic string detection model of the photovoltaic power station DOM image is used to detect the photovoltaic panel strings in the ground DOM image of the photovoltaic power station, specifically including: S21. Collect ground DOM images of photovoltaic stations and build a training dataset; S22. Build a photovoltaic string detection model using a deep learning algorithm. The model input is a DOM image, and the output is the location and bounding box of the photovoltaic string. S23, labeling the photovoltaic strings in the training data set using a labeling tool; S24. Training the photovoltaic string detection model using the training data set, optimizing the model parameters, and improving the detection accuracy; S25, inputting the ground DOM image into the trained photovoltaic string detection model, and the model detects and outputs the location and bounding box information of the photovoltaic strings; S26. Post-process the detection results and remove redundant detection frames with high overlap through non-maximum suppression algorithms, such as NMS or Soft-NMS algorithms, to reduce false detections. Furthermore, a confidence threshold can be set to filter out low-confidence detection results, further reducing the false detection rate, removing falsely detected and missed photovoltaic strings, and ensuring the accuracy of the detection results. The obtained bounding box is mapped to an image mask to obtain a mask image of the photovoltaic strings in the photovoltaic station, such as Figure 3 shown.

[0034] S3. Construct a walkable maintenance area around the detected PV strings and obtain the skeleton of the walkable path and the connectivity between the skeleton nodes; specifically, the following steps are performed: S31, merge adjacent photovoltaic strings, fill the gaps between the photovoltaic strings, and expand the area around the photovoltaic strings as a walkable area, such as Figure 4 As shown; Preferably, the merging process includes: performing a morphological closing operation on adjacent strings with a spacing of less than 1.5 m, with a kernel size of 15×15 pixels; The expanded area around the photovoltaic string as the walkable area includes: a walkable area surrounding the photovoltaic string generated by expanding 2 meters (DOM pixel width is about 60pix) along the string boundary.

[0035] S32, refine the walkable area to obtain a two-dimensional skeleton image of the walking path. The pixel width of the path skeleton is 1, and the connection between each skeleton segment is a skeleton node, such as Figure 5 As shown; Preferably, the walkable area is thinned: the Zhang-Suen parallel thinning algorithm is used, iterated 12 times to obtain a skeleton image, 386 nodes and 452 edges are extracted, and an adjacency matrix is constructed to store connectivity relationships; S33. Detect bifurcation nodes in the refined skeleton image and calculate connectivity relationships between the nodes.

[0036] S4. Determine the location of the fault point in the walkable maintenance area, and plan the shortest maintenance path of the photovoltaic power station based on the shortest path algorithm, such as the Dijkstra algorithm, the Floyd-Warshall algorithm; Take the points on the maintenance path in front of the faulty photovoltaic string as maintenance points, and detect the connectivity between the skeleton nodes. Are the starting point and the ending point specified? Local shortest path planning: When specifying the starting maintenance point, adopt local graph theory modeling, use methods such as the nearest neighbor algorithm to solve or TSP to solve, and combine dynamic adjustment strategies to generate the local shortest maintenance path and plan the local shortest maintenance path of the photovoltaic power station; The idea of graph theory modeling is: represent the factors of the model and their connections in the language of graph theory; transform the actual problem into a graph theory problem, such as the adjacency matrix representation method, the adjacency list representation method, etc. of the graph; The nearest neighbor algorithm is the KNN algorithm. The above local graph theory modeling, the nearest neighbor algorithm to solve or TSP to solve methods and dynamic adjustment strategies can all be implemented by those skilled in the art according to the algorithm name to realize the relevant algorithms and functions.

[0037] Global shortest path planning: When not specifying the starting maintenance point, adopt methods such as global graph theory modeling, the nearest neighbor algorithm to solve or TSP to solve, use multi-objective optimization and path segmentation methods, and combine dynamic update and parallelization strategies to generate the global shortest maintenance path and plan the global shortest maintenance path of the photovoltaic power station.

[0038] The above global graph theory modeling, the nearest neighbor algorithm to solve or TSP to solve methods, multi-objective optimization and path segmentation methods, dynamic update and parallelization strategies can all be implemented by those skilled in the art according to the algorithm name to realize the relevant algorithms and functions.

[0039] Preferably, input the coordinates of 5 fault points, adopt the Dijkstra algorithm combined with TSP constraints, the initial path: A→D→B→E→C, the total length is 150 meters, and the path planning takes 0.23 seconds, which is a significant improvement compared to the 67 seconds consumed by the traditional A* algorithm. As Figure 6 shown.

[0040] S5. Optimize the shortest maintenance path based on the node stretching algorithm: Determine the nodes passed through in the planned maintenance path and the pixel lengths between adjacent nodes, stretch and optimize the paths between adjacent nodes, and deploy them to the terminal hardware device.

[0041] Preferably, determine the nodes passed through in the planned maintenance path and the pixel lengths between adjacent nodes, and stretch and optimize the paths between adjacent nodes. Specifically: S51. Set the starting point of the path as point P; S52. Traverse pixel points C passed by the maintenance path after point P one by one, and connect pixel point C and point P with a straight line; S53. Check whether the connected straight line crosses the mask area of the photovoltaic string; S54. If it crosses, stop traversing, use the straight line between point P and point C as the optimized path after stretching between the two points, update this pixel point C to P, and continue to iterate steps S52 and S53; S55. If it does not cross, continue to iterate steps S52 and S53; S56. If pixel C is the end point of the path, iteratively execute steps S51 - S55 until the length of the optimized maintenance path does not change, and the obtained optimized maintenance path is the optimal maintenance path, as Figure 6 shown; S57. Convert the DOM image pixels of each node in the optimal maintenance path into ground longitude and latitude coordinates according to the engineering coordinate system stored in the DOM image, which can provide navigation guidance for the fault string detection of the photovoltaic power station.

[0042] Preferably, iteratively optimize the initial path: The 1st iteration: The path is shortened to 115 meters, reducing by 35 meters; The 3rd iteration converges: The final path length is 108 meters, as Figure 5 shown, shortened by 28% compared with before optimization; Key optimization section: Detection of the straight line crossing of section D→E, with a cumulative shortening of 10 meters.

[0043] Preferably, the deployment to the terminal hardware device includes: First, export the photovoltaic string detection model in the method as an onnx format model, and use the optimization tool of ONNX (such as onnx - simplifier) to simplify the exported ONNX model, remove redundant operations, and reduce the model complexity; Secondly, use RKNN Toolkit to load the optimized ONNX model, perform quantization processing on the ONNX model, and convert the floating - point model into a fixed - point rknn model, aiming to improve the inference speed of the model on the RV1126 terminal device; Then, deploy the generated rknn model to the target hardware platform, and further optimize the model or hardware configuration according to the actual running performance (such as inference speed, memory occupancy, etc.), which can be optimized or set by those skilled in the art; Finally, convert other parts in the method into a terminal program, deploy it to the RV1126 platform, and achieve autonomous maintenance path planning on the terminal platform.

[0044] Preferably, deploy on the Rockchip RV1126 platform: Model Quantization: FP32 → INT8, model size compressed from 86 MB to 23 MB; Memory Occupancy: Peak memory < 512 MB; End-to-End Latency: The entire process from DOM input to path output < 8 seconds.

[0045] Embodiment 2 This embodiment provides a device for implementing a method for planning maintenance paths of a photovoltaic power station. The device includes: At least one processor; And a memory that stores instructions. When the instructions are executed by the at least one processor, the at least one processor executes a method for planning maintenance paths of a photovoltaic power station as described above.

[0046] In this embodiment, the electronic device includes but is not limited to: personal computers, server computers, workstations, desktop computers, laptop computers, notebook computers, mobile computing devices, smart phones, tablet computers, cellular phones, personal digital assistants (PDAs), handheld devices, messaging devices, wearable computing devices, consumer electronic devices, etc.

[0047] Embodiment 3 This embodiment also provides a computer-readable storage medium that stores executable instructions. When the instructions are executed, the machine executes a method for planning maintenance paths of a photovoltaic power station as described above.

[0048] Specifically, a system or device equipped with a readable storage medium can be provided. On this readable storage medium, software program code for implementing the functions of any one of the above embodiments is stored, and the computer or processor of the system or device reads and executes the instructions stored in the readable storage medium.

[0049] In this case, the program code read from the readable medium itself can implement the functions of any one of the above embodiments. Therefore, the computer-readable code and the readable storage medium storing the computer-readable code constitute a part of this specification.

[0050] Examples of readable storage media include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD-RW), magnetic tapes, non-volatile memory cards, and ROMs. Optionally, the program code can be downloaded from a server computer or cloud via a communication network.

[0051] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0052] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0053] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that realizes the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0054] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0055] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the technical solutions of the present invention, rather than limitations on the specific implementation manners of the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the claims of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. A method for planning the maintenance path of a photovoltaic power station, characterized in that, The method includes: S1. Collect the orthophoto images of the photovoltaic power station and synthesize the ground DOM image of the photovoltaic power station; S2. Detect the photovoltaic string in the ground DOM image of the photovoltaic power station through the photovoltaic string detection model of the photovoltaic power station DOM image; S3. Construct a walkable maintenance area around the detected photovoltaic string, and obtain the skeleton of the walkable path and the connectivity relationship between the skeleton nodes; S4. Determine the position where the fault point is located in the walkable maintenance area, and plan the shortest maintenance path of the photovoltaic power station based on the shortest path algorithm; S5. Determine the nodes passed through in the planned maintenance path and the pixel length between adjacent nodes, stretch and optimize the path between adjacent nodes, and deploy it to the terminal hardware device.

2. The photovoltaic power station maintenance path planning method according to claim 1, wherein The specific content of S2 includes: S21. Collect the ground DOM images of the photovoltaic power station and construct a training data set; S22. Use a deep learning algorithm to construct a photovoltaic string detection model; S23. Annotate the photovoltaic strings in the training data set through an annotation tool; S24. Train the photovoltaic string detection model with the training data set to optimize the model parameters; S25. Input the ground DOM image into the trained photovoltaic string detection model, and the model detects and outputs the position and bounding box information of the photovoltaic string; S26. Perform post-processing on the detection results. Remove redundant detection frames with high overlap through the non-maximum suppression algorithm to reduce false detections; at the same time, set a confidence threshold to filter out detection results with low confidence; then, map the obtained bounding boxes to an image mask to obtain the photovoltaic string mask image in the photovoltaic power station.

3. A photovoltaic power station maintenance path planning method according to claim 1 or 2, characterized in that, The specific content of S3 includes: S31. Perform a merging process on adjacent photovoltaic strings, fill the gaps between the photovoltaic strings, and expand the area around the photovoltaic strings as a walkable area; S32. Refine the walkable area to obtain a two-dimensional skeleton image of the walking path. The pixel width of the path skeleton is 1, and the connection points of each section of the skeleton are skeleton nodes; S33. Detect bifurcation nodes in the refined skeleton image and calculate the connectivity relationship between the nodes.

4. A method for planning the maintenance path of a photovoltaic power station according to claim 3, characterized in that, The merging process includes: performing a morphological closing operation on adjacent strings with a spacing <1.5m, and the kernel size is 15×15 pixels; The expansion of the area around the photovoltaic string as a walkable area includes: expanding 2 meters outward along the string boundary to generate a walkable area surrounding the photovoltaic string.

5. A photovoltaic power station maintenance path planning method according to claim 4, characterized in that, The refinement process of the walkable area includes: using the Zhang-Suen parallel refinement algorithm, iterating 12 times to obtain a skeleton image, extracting 386 nodes and 452 edges, and constructing an adjacency matrix to store the connectivity relationship.

6. A photovoltaic power station maintenance path planning method according to claim 5, characterized in that The specific content of S4 includes: Take the points on the maintenance path in front of the faulty photovoltaic string as maintenance points; When specifying the starting maintenance point, use local graph theory modeling, use the nearest neighbor algorithm to solve combined with a dynamic adjustment strategy to generate a local shortest maintenance path, and plan the local shortest maintenance path of the photovoltaic power station; When not specifying the starting maintenance point, use global graph theory modeling, the nearest neighbor algorithm to solve, and at the same time use multi-objective optimization and path segmentation methods, combined with dynamic update and parallelization strategies to generate a global shortest maintenance path, and plan the global shortest maintenance path of the photovoltaic power station.

7. A method for planning a maintenance path of a photovoltaic power station according to claim 6, characterized in that, Determine the nodes passed through in the planned maintenance path and the pixel length between adjacent nodes, and stretch and optimize the path between adjacent nodes, specifically including: S51. Set the starting point of the path as point P; S52. Traverse the pixel points C passed through by the maintenance path after point P one by one, and connect the pixel points C and point P with a straight line; S53. Check whether the connected straight line crosses the mask area of the photovoltaic string; S54. If it crosses, stop traversing, take the straight line between point P and point C as the optimized path after stretching between the two points, update the pixel point C to P, and continue to iterate steps S52 and S53; S55. If it does not cross, continue to iterate steps S52 and S53; S56. If pixel C is the end point of the path, iterate and execute steps S51 - S55 until the length of the optimized maintenance path does not change. The obtained optimized maintenance path is the optimal maintenance path; S57. Convert the DOM image pixels of each node in the optimal maintenance path into ground longitude and latitude coordinates according to the engineering coordinate system stored in the DOM image, so as to provide navigation guidance for the fault string detection of the photovoltaic power station.

8. A method for planning a maintenance path of a photovoltaic power station according to claim 7, characterized in that, The deployment to the terminal hardware device includes: First, export the photovoltaic string detection model as an onnx format model, and use the optimization tool of ONNX to simplify the exported ONNX model; Second, use RKNN Toolkit to load the optimized ONNX model, perform quantization processing on the ONNX model, and convert the floating-point model into a fixed-point rknn model; Then, deploy the generated rknn model to the target hardware platform, and optimize the model or hardware configuration according to the actual operation performance; Finally, convert the other parts of the method into a terminal program and deploy it to the RV1126 platform to realize autonomous maintenance path planning on the terminal platform.

9. An apparatus for a photovoltaic power station maintenance path planning method, characterized in that, The device includes: A processor; A memory, on which a computer program that can run on the processor is stored; Wherein, when the computer program is executed by the processor, it implements the steps of a method for planning a maintenance path of a photovoltaic power station as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method as described in any one of claims 1 to 8.

Images