A method and system for monitoring and analyzing power cables

Abstract

The present application relates to the technical field of power inspection, and discloses a power cable monitoring and analysis method and system thereof. The method includes obtaining the overall topological network of the power system and the number of unmanned aerial vehicles (UAVs), partitioning the nodes of the overall topological network to obtain a plurality of local topological networks; creating inspection tasks and task material information based on the local topological networks and distributing them to each UAV respectively; controlling the UAVs to go to the starting points of the corresponding partitions, traversing each node and branch based on a preset rule and recording, and inspecting the wires during the traversal; controlling the UAVs to generate a new partition topological network based on the recorded nodes and branches; after all the UAVs complete the inspection tasks and return, obtaining the nodes and branches recorded by the UAVs, summarizing the partition networks into a total topological network, and deleting the repeatedly generated branches and nodes. The present application has the advantages of reducing the problems of missed inspection, duplicate inspection, and repeated path flight caused by positioning drift or line changes.

Description

Technical Field

[0001] This application relates to the technical field of power line inspection, and in particular, to a method and system for monitoring and analyzing power cables. Background Art

[0002] The power system plays a crucial role in the infrastructure construction of modern society, meeting the electricity demands of cities, industries, and remote areas by transmitting high-voltage electric energy. Since transmission lines often span large geographical areas and are distributed in high or complex terrains, the maintenance and fault detection of power lines are particularly important. If potential risks such as wire slack, insulation damage, pole tilt, and foreign object entanglement are not detected and eliminated in a timely manner, they may lead to large-scale power outages, safety accidents, or economic losses; therefore, maintaining the real-time healthy state of transmission lines has always been a key task faced by power grid enterprises.

[0003] To improve the inspection efficiency, existing technologies usually use drones equipped with cameras or sensors to visually inspect power lines and utility poles. A common practice is to rely on GPS and geographical coordinates to fly the drone along a known route autonomously or semi-autonomously, and then the captured video and image data are used by maintenance personnel or recognition algorithms for fault detection. Although GPS is convenient for large-scale positioning, it often faces various challenges in the actual application scenarios of the power transmission network: on the one hand, the distance between utility poles may be relatively short, and the accuracy and stability of GPS signals are difficult to meet the precise navigation requirements of several meters or even sub-meter levels; on the other hand, in the face of surrounding buildings, terrain occlusion, or signal interference, the drift phenomenon of GPS will be more obvious. In addition, the power network is constantly changing: new utility poles may be built, some old lines may be removed, or some branches may be temporarily disconnected for maintenance. Coupled with situations such as cable sag and arc change, the method of simply relying on GPS path inspection will result in missed inspections or repeated flights. Summary of the Invention

[0004] To ensure the comprehensiveness and efficiency of the inspection and reduce the problems of missed inspections, duplicate inspections, and repeated path flights caused by positioning drift or line changes, this application provides a method and system for monitoring and analyzing power cables.

[0005] In a first aspect, this application provides a method for monitoring and analyzing power cables, adopting the following technical solution:

[0006] A method for monitoring and analyzing power cables includes the following steps:

[0007] S1. Obtain the overall topological network of the power system and the number of drones, and partition the nodes of the overall topological network based on the number of drones to obtain multiple local topological networks. Herein, the overall topological network includes nodes and branches, the nodes correspond to utility poles, and the branches correspond to wire branches;

[0008] S2. Create inspection tasks and task material information based on the local topological networks and send them to each drone respectively. The task material information includes the node information and branch information corresponding to a certain local topological network, as well as the pre-entered node image information corresponding to each node. The edge of the local topological network is a node, and adjacent local topological networks are connected by shared nodes;

[0009] S3. Control the drones to go to the starting point of the corresponding partition, and then start from the starting point, traverse each node and branch based on preset rules and record, and check the wires during the traversal;

[0010] S4. Control the drones to generate a new partition topological network based on the recorded nodes and branches;

[0011] S5. After all drones complete the inspection tasks and return, obtain the nodes and branches recorded by the drones, determine the common intersection points generated by drones in adjacent areas, summarize the partition networks into a total topological network, and delete the repeatedly generated branches and nodes.

[0012] Optionally, S1 includes the following steps:

[0013] S11. Obtain the overall topological network of the power system and the number N of drones, and determine the distribution range of the number of branches and the distribution range of the number of nodes of the local topological network based on the total number of nodes, the total number of branches of the overall topological network, and the number of drones;

[0014] S12. Make a preliminary judgment and screening on each node of the overall topological network according to the branch quantity and distance constraints, and select N candidate nodes as the possible centers of the local topological network;

[0015] S13. With each candidate center point as the core, absorb the unowned nodes into their respective sub-networks layer by layer outward or in the shortest path manner, where the absorption speed of each candidate center is the same;

[0016] S14. When the number of nodes or the number of branches in a certain sub-network reaches the upper limit, stop absorbing new nodes into this sub-network;

[0017] S15. Allocate the remaining nodes without attribution to avoid the appearance of isolated nodes or redundant sub-networks;

[0018] S16. Inside each local topology network, check the path length from the central point to each edge node. If it exceeds or is lower than the preset threshold, then by means of boundary node adjustment, reassign some nodes to adjacent local topology networks so that the distance from the central point to the edge gradually returns to a reasonable range;

[0019] S17. Re-evaluate the position of the central point so that the difference between the maximum node distance and the minimum node distance from the central point to the edge in the local topology network is less than the preset threshold; wherein, the position of the central point serves as the starting point of the drone's flight in the partition;

[0020] S18. Check the fulfillment of the constraints of the local topology network and perform step backtracking for adjustment.

[0021] Optionally, the S3 includes:

[0022] S31. The drone goes to the starting point of the corresponding partition;

[0023] S32. The drone flies to the reference point corresponding to the utility pole based on image recognition;

[0024] S33. The drone creates a blank new topology network diagram and generates nodes corresponding to the utility pole where it is located within the new topology network diagram;

[0025] S34. The drone adjusts to a preset attitude at the reference point and takes a photo at the node position to obtain a node position image;

[0026] S35. The drone identifies the wires in the node position image with the pre-entered node image information as a prior condition;

[0027] S36. The drone generates branches corresponding to the wires based on the node position image and generates nodes corresponding to the utility pole where it is located within the new topology network diagram;

[0028] S37. The drone determines the branch inspection order of the drone at this node based on the new topology network diagram and checks the wires corresponding to the branches according to the branch inspection order and inspection strategy.

[0029] Optionally, the inspection strategy includes:

[0030] The drone preferentially checks the branches according to the branch inspection order set by the local topology network at this node. If there are still branches that have not been inspected after the inspection of the wires corresponding to the local topology network in the new topology network diagram, then inspect these branches;

[0031] The drone determines whether the node reached during branch inspection is the corresponding node within the local topology network based on image recognition. If so, it determines whether all the branches corresponding to this node have been traversed. If not, it proceeds to the next branch according to the branch inspection order. If so, it returns to the previous node.

[0032] Otherwise, it calculates the distance between this node and the center point. If this distance is less than or equal to the distance to the edge, it determines that this node is a new node in this partition. If it is greater, it controls the drone to return to the previous node.

[0033] Optionally, S32 includes the following sub-steps:

[0034] S321. When the drone reaches the area corresponding to the starting point, it acquires an image of the utility pole at the node.

[0035] S322. Based on image recognition, the drone determines the position of the utility pole in the image.

[0036] S323. The drone selects a position at a preset height above the utility pole as the reference point.

[0037] S324. The drone flies to the reference point.

[0038] Optionally, S35 includes the following steps:

[0039] S351. Acquire pre-entered node image information. Among them, the pre-entered node image contains a utility pole and multiple wires, and the pre-entered node image is taken at the reference point corresponding to the utility pole in a preset pose.

[0040] S352. Acquire the node position image, where the node position image is taken at the reference point corresponding to the utility pole in a preset pose.

[0041] S353. Register the node position image and the pre-entered node image.

[0042] S354. Perform wire detection on the node position image to identify slender lines or arcs.

[0043] S355. Perform feature analysis on the detected line segments and merge approximately continuous short line segments to obtain the wire contour. Among them, the feature analysis includes length analysis, angle analysis, and continuity analysis.

[0044] S356. In the registered coordinate system, compare the starting point, ending point, direction, and length of the line segments in the two photos. When the similarity or matching degree meets the set threshold, it is determined to be the same wire. Among them, the matching degree includes endpoint distance, average offset, and angle difference.

[0045] If no matching object is found or the matching degree is low, it is determined that a new wire is added or a wire is reduced; where a new wire corresponds to a wire that exists in the node position image but does not exist in the pre-entered node image; a reduced wire corresponds to a wire that does not exist in the node position image but exists in the pre-entered node image.

[0046] Optionally, S37 includes the following steps:

[0047] S371. Control the drone to obtain image information and obtain wire image information based on instance segmentation;

[0048] S372. Input it into a pre-trained deep learning model to identify the risk type;

[0049] S373. If a risk is identified, record the branch and mark the risk type, and take a photo of the risk location.

[0050] Among them, the training steps of the pre-trained deep learning model include:

[0051] S3721. Determine the risk types to be identified and define the recognizable features of each risk type;

[0052] S3722. Obtain image samples containing risk recognizable features of various types, and manually mark the recognizable features in the images using annotation tools;

[0053] S3723. Divide the image samples into a training set and a test set based on a pre-selected partitioning strategy and partitioning ratio. Among them, the training set is used to train the model, and the test set is used to verify and adjust the model. The pre-selected partitioning strategy is random partitioning, stratified partitioning, or using cross-validation;

[0054] S3724. Train a CNN model using the ResNet architecture based on the training set, and use the cross-entropy loss function during the training process;

[0055] S3725. Input the test set into the trained model, input the risk types identified by the trained model into a preset reward function for calculation and obtain a reward value. Among them, the output of the preset reward function is related to the difference between the risk types identified by the CNN model and the actual risk types. When the difference is larger, the reward value is lower, and when the difference is smaller, the reward value is smaller;

[0056] S3726. Adjust the parameters of the trained model based on the reinforcement learning algorithm and the change of the reward value to increase the reward value, so as to obtain new risk types, substitute them into the previous step and execute until the risk types converge to a stable range.

[0057] In a second aspect, the present application provides a power cable monitoring and analysis system, adopting the following technical solutions:

[0058] A power cable monitoring and analysis system includes a processor, and a program of the power cable monitoring and analysis method described in any one of the above is run in the processor.

[0059] In a third aspect, the present application provides a storage medium, adopting the following technical solution:

[0060] A storage medium stores a program of the power cable monitoring and analysis method described in any one of the above.

[0061] In summary, the present application includes at least one of the following beneficial technical effects: It can avoid over-reliance on GPS, and can also quickly locate and identify the increased and decreased nodes and branches in a dynamic power system, realizing the timely discovery and maintenance of potential faults. By combining the prior image information taken by the unmanned aerial vehicle at each node on the basis of the original topology map of the power network, the increased and decreased wires and nodes can be locally updated, and finally an accurate and up-to-date overall topology network can be formed after the data of multiple unmanned aerial vehicles are summarized, thus ensuring the comprehensiveness and efficiency of the inspection, and reducing the problems of missed inspection, duplicate inspection, and repeated path flight caused by positioning drift or line changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 is a flowchart of the power cable monitoring and analysis method in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0063] The following details the embodiments of the present application, and the examples of the embodiments are shown in the drawings.

[0064] In the description of this specification, the description referring to the terms "certain embodiments", "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0065] The embodiment of the present application discloses a power cable monitoring and analysis method, referring to Figure 1 , the method includes the following steps S1-S5.

[0066] S1. Obtain the overall topological network of the power system and the number of drones, and partition the nodes of the overall topological network based on the number of drones to obtain multiple local topological networks. Here, the overall topological network includes nodes and branches, where the nodes correspond to utility poles and the branches correspond to wire branches.

[0067] In this step, first, it is necessary to obtain the overall topological network of the power system and the number of drones to clarify all the nodes involved in the inspection process and the equipment resources available for inspection. The "overall topological network" mentioned here refers to a complete graphical or data structure composed of utility poles and wire branches. Among them, the "nodes" correspond to utility poles, which undertake the functions of supporting power transmission or information intersection; the "branches" correspond to wire branches, which connect each utility pole and form a complete power supply path. By mastering the global information of these nodes and branches, more precise division of labor can be carried out in subsequent drone scheduling and inspection route planning. For example, if there are a total of 100 utility poles and 200 wire branches in an area, and only 5 drones are available, then these 100 nodes need to be reasonably divided into 5 groups so that each drone can concentrate and efficiently complete the inspection of its respective responsible area.

[0068] After obtaining the overall network and the number of drones, the nodes are partitioned according to the number of drones, splitting the large power network into several smaller and more manageable and inspectable local topological networks. The purpose of this is to reduce the operation burden of a single drone, enable each drone to have a relatively balanced inspection range, and at the same time reduce the resource waste caused by repeated flights or mutual interference between drones in a complex environment. Through this partitioning method, on the one hand, it can ensure that all nodes can be included in the inspection range, reducing omissions and inspection blind spots; on the other hand, it is easier to locate specific branches or nodes in case of a failure, which is very convenient for subsequent maintenance and correction.

[0069] Specifically, S1 includes the following steps S11 - S18.

[0070] S11. Obtain the overall topological network of the power system and the number of drones N, and determine the distribution range of the number of branches and the distribution range of the number of nodes of the local topological network based on the total number of nodes, the total number of branches of the overall topological network, and the number of drones.

[0071] In this step, the system first obtains all the node and branch information that constitutes the overall power system, that is, collects the connection relationships between each utility pole and wire branch, and then determines the number N of drones available for inspection. By mastering these three key data, namely the total number of nodes, the total number of branches, and the number of drones, the system can formulate a reasonable inspection area division plan at the planning level. Here, "node" refers to the utility pole, and "branch" refers to the wire branch connecting the utility poles. For example, if a certain power system has 150 nodes, 280 branches, and 4 available drones, the system will comprehensively consider the number of nodes and branches that each local network should contain and determine an allocation range. For instance, at the above scale, each drone can be responsible for 30 to 45 nodes and 50 to 90 branches, which can ensure that each local topology network has a relatively balanced workload, avoiding overloading a single drone while also guaranteeing the overall inspection efficiency of the system.

[0072] S12. According to the branch quantity and distance constraints, conduct a preliminary judgment and screening on each node of the overall topology network, and select N candidate nodes as the possible centers of the local topology network.

[0073] In this step, the system will conduct a preliminary screening on the nodes corresponding to each utility pole within the overall topology network according to the branch quantity and distance constraints. The goal is to select suitable "center" nodes from numerous nodes for subsequent formation of the core of the local topology network. In this process, it is necessary to comprehensively evaluate the branch quantity, connectivity, and distance distribution from this node to other nodes around the node. For example, if the distribution of utility poles in a certain area is very dense and the distance between some nodes is relatively far, it is necessary to give priority to those nodes that are relatively balanced with nearby nodes and where situations of being too far or too close will not occur over a large area. Only in this way can it be ensured that in the subsequent inspection process, each local network will not be too concentrated or too dispersed.

[0074] When performing the above screening, the system often uses certain algorithms to calculate the average distance, maximum distance, and minimum distance between each node and its neighboring nodes to determine whether they meet the pre-set thresholds. The finally selected N candidate nodes are regarded as having the potential to become the centers of the local network. Their positions can usually cover a certain range of utility poles and prevent the number of nodes and branches within the local network from being difficult to manage due to over-concentration.

[0075] S13. With each candidate center point as the core, absorb the unowned nodes into their respective sub-networks layer by layer outward or in the shortest path manner, where the absorption speed of each candidate center is the same.

[0076] In this step, each selected candidate center point will serve as the "core" of a local topological network, gradually absorbing the nodes in the surrounding area that have not yet been assigned to any sub-network. The so-called "layer-by-layer" method means that it can start from the central node and radiate outwards to the adjacent nodes surrounding it; while the "shortest path" method is to classify other nodes under the nearest center point according to the connection or path length between nodes, following the principle of the shortest distance or the fewest hops. Unowned nodes refer to those that have not been included in any local network before, and they will be gradually assigned to the more suitable or optimal center point during this process. To prevent some center points from "grabbing" too many nodes too quickly, the absorption speed of all candidate centers will be ensured to be the same, so as to ensure that the division process of the entire topological network is relatively balanced and there will be no problem of excessive expansion of individual sub-networks. For example, in a certain area, if center point A and center point B are not far apart, and there are several nodes that have short paths to both of them at the same time, then the system will calculate the shortest paths of each node to A and B for comparison, so as to decide which center point to assign the node to. The purpose of doing this is to keep the structure of each sub-network stable and reasonably distributed, which can not only reduce the overlap of the subsequent inspection ranges of the drones, but also improve the overall inspection efficiency and coverage.

[0077] S14. When the number of nodes or the number of branches in a sub-network reaches the upper limit, stop absorbing new nodes into this sub-network.

[0078] In this step, when the number of nodes or the number of branches in a sub-network reaches the pre-defined upper limit, the system will temporarily stop incorporating new nodes into this sub-network to prevent it from expanding excessively. The "upper limit" mentioned here usually refers to the reasonable load range set for each local topological network before. For example, when calculating the number of nodes and the number of branches in the local topological network, the inspection capabilities and inspection time of the drones have been estimated. If the burden of any sub-network exceeds this range, it will bring difficulties or unfair distribution to the subsequent actual inspections.

[0079] In the specific implementation, the system will continuously monitor the number of nodes and the number of branches in the current sub-network during the network expansion process. Once it detects that a certain sub-network reaches the preset upper limit, it will mark this sub-network as "fully loaded" and prevent new nodes from joining. This approach can be vividly compared to "stopping when full": when the serviceable range of a certain center point is full, no more remaining nodes will be accepted, thus preventing unbalanced situations. For example, if a sub-network has already accommodated 40 nodes and its threshold is exactly 40, the system will interrupt the expansion of this center and require the remaining unassigned nodes to seek other center points, thus maintaining a relatively balanced scale among the sub-networks.

[0080] S15. Allocate the remaining unassigned nodes to avoid the emergence of isolated nodes or redundant sub-networks.

[0081] During this process, the system will re-allocate the remaining nodes that have not been absorbed by any sub-network to ensure the integrity of the entire network. The emergence of "remaining nodes" is often due to the fact that some nodes were not divided in a timely manner due to distance or the number of branches in the previous steps, or the neighboring central points have reached the upper limit and stopped absorbing. By re-assigning these "ownerless" nodes to appropriate sub-networks, the phenomenon of "isolated nodes" or the formation of "extra small networks" privately in the topological structure can be avoided, thus maintaining the order and coherence of the overall division. For example, if there are several scattered utility poles in a specific area that are rejected by the neighboring sub-networks (because the neighboring sub-networks are saturated), then the system needs to identify these nodes and assign them to other central points that are the closest and still have capacity. This not only ensures that all utility poles can be included in the inspection coverage, but also avoids the embarrassing situation of subsequent drone scheduling difficulties. The final effect is to keep the number of local topological networks always consistent with the previous design goal, reducing the risk of duplication or omission in inspection and management.

[0082] Alternatively, in some other embodiments, when the system finds that there are remaining nodes to be allocated but are adjacent to a sub-network that has reached its capacity limit, this problem can be solved through a "node replacement" method. Specifically, first temporarily assign the remaining node to the adjacent saturated sub-network, and then select a node to "release" from the junction of this sub-network and other sub-networks that still have capacity, and transfer it to an adjacent sub-network that still has capacity. This seemingly adds a new node to the originally saturated sub-network, but in fact, by transferring a node to the nearby sub-network, it ensures the overall network connectivity and load balance. For example, if a sub-network has reached its node limit because it has absorbed a large number of utility poles during the previous division, but is adjacent to a sub-network that still has expandable space, the newly arrived remaining node can be first placed in this full sub-network, and then a suitable node for transfer can be selected from the edge between it and the idle sub-network for exchange. This not only enables both sub-networks to have a coherent utility pole layout, but also ensures that there are no isolated or redundant nodes, while maintaining the overall connectivity of the network structure both physically and logically.

[0083] S16. Inside each local topological network, check the path length from the central point to each edge node. If it exceeds or is lower than the preset threshold, then through the method of adjusting the boundary nodes, re-assign some nodes to the adjacent local topological network so that the distance from the central point to the edge gradually returns to a reasonable range.

[0084] In this step, the system will further examine the node distribution within each local topology network, focusing on whether the path length from the central point to the edge nodes exceeds or falls below a predefined threshold. Here, the "central point" refers to the core node selected during the previous screening or calculation of the local topology network, while the "edge nodes" are the nodes distributed at the outermost periphery of this sub-network, which may be connected to other sub-networks or are relatively far from the center. The system will calculate the shortest path or average path distance from the central point to the edge nodes. If these values deviate from the preset upper and lower limits, it indicates that the sub-network is either too large or too small, or there is an unreasonable node allocation at the boundary, and appropriate regulation is required. It should be noted that the reasonable range mentioned in S16 is the average value of the distances from the central points to the edge nodes of each local topology network ±1. Of course, in other embodiments, it can also be other values.

[0085] In specific implementation, usually the non-compliant edge nodes and their related connections will be re-assigned to adjacent sub-networks. Through this method of "boundary node adjustment", the nodes that were originally too far or too close to each other will be attributed to a more appropriate central point. For example, if the center of a certain sub-network is located in an area with relatively dense power poles, but there are still a few nodes at its outermost periphery that are too far from the center and are not convenient for inspection and scheduling, these edge nodes can be transferred to the adjacent sub-network, so that all sub-networks are more balanced in terms of geographical distribution and connection relationship. The purpose of doing this is to maintain the compactness and manageability within each sub-network, avoid the drone spending too much flight time in a single area due to excessive distance during subsequent inspection, and at the same time reduce potential conflicts or resource waste caused by excessive overlap between sub-networks.

[0086] S17. Re-evaluate the position of the central point so that the difference between the maximum node distance and the minimum node distance from the central point to the edge in the local topology network is less than a preset threshold; where the position of the central point serves as the starting point for the drone's flight in the partition.

[0087] The system will re-evaluate the position of the center point of the local topology network so that the difference between the maximum distance and the minimum distance from this center point to the edge nodes of the network is less than a preset threshold. The center point is the most crucial reference position of a local network and also the starting point for the UAV to perform subsequent inspection tasks. If the gap between the farthest and the nearest nodes is too large, it indicates that the layout inside the network is not balanced enough, which may lead to inefficiencies or uneven coverage during the flight and inspection of the UAV. To solve this problem, the system will calculate the actual distance distribution from the current center point to all edge nodes and make the edge distances more evenly distributed by replacing or adjusting the center point. For example, in a set of nodes distributed in an irregular shape, if the initially selected center point is biased towards one side of the network, then some edge nodes in the diagonal position will appear overly distant. By moving the center point or reselecting a node closer to the overall geometric center, the distance to the far end can be effectively reduced, thereby keeping the distance distribution within a reasonable range.

[0088] During the implementation process, the system will combine the analysis results of the shortest path or average distance between nodes, and try to re-correct the center points of sub-networks with too large distance differences to a more balanced position, or select a pole that is more evenly distributed with edge nodes as the new core node. This not only makes the structure of the local topology network more compact, but also makes the inspection range more reasonable, and can improve the subsequent path planning effect. When the UAV departs from this "updated" center point, whether it flies to the surrounding close-range nodes or covers the more distant nodes, it can better complete the inspection within the established flight range and time limit.

[0089] S18. Check the constraint achievement of the local topology network and perform step back for adjustment.

[0090] At this stage, the system will conduct an overall verification of all the constraint conditions of the local topological network to determine whether the set values are achieved in terms of the number of nodes, the number of branches, the distance range between the central point and the edge nodes, etc. If it is found during the inspection that there are still sub-networks that do not meet the requirements, or if the node allocation in some areas is unreasonable, it will roll back to the corresponding steps according to the type of problem and re-execute the fine-tuning. For example, if the system detects that the number of nodes in a sub-network exceeds the upper limit, it will roll back to the previous link responsible for network absorption and node movement; if it is found that the distances from the central point to some edge nodes are still unbalanced, it will roll back to the steps to adjust the position of the central point or the allocation strategy. Through this iterative method, whether it is node allocation, branch division, or distance correction, multiple opportunities for correction can be obtained to gradually approach a balanced local topological structure that meets the requirements of the inspection tour. For example, if a sub-network is overloaded due to accidentally absorbing several more nodes, some nodes can be transferred by rolling back to the previous absorption or release process, or the problem of excessive distances of remote nodes can be alleviated by further adjusting the central point.

[0091] S2. Create inspection tour tasks and task material information based on the local topological network and distribute them to each UAV respectively. Among them, the task material information includes the node information and branch information corresponding to a certain local topological network, as well as the pre-entered node image information corresponding to each node. The edge of the local topological network is a node, and adjacent local topological networks are connected through shared nodes.

[0092] The system will generate inspection tour tasks and the required material information for each UAV respectively based on the previously divided local topological network. The so-called "local topological network" refers to splitting the entire power system into several relatively independent network areas, each area containing several utility poles (i.e., nodes) and the wire branches between them. These information have been clarified during the zoning process in the previous stage. When creating the inspection tour tasks, it is necessary to collate the node information, branch information corresponding to the local topological network, and the pre-entered node images (i.e., pre-stored or captured image materials) corresponding to each node to form "task materials". In this way, each UAV can accurately locate utility poles, branches, and other key elements by combining the graphic and text materials of the local area after receiving the distributed tasks.

[0093] In a more specific implementation, the system determines a flight coverage range for each local topology network and indicates all the utility poles included therein and the wire routing between the utility poles in the task data. For example, if a local topology network includes ten nodes, several branches, and one or more pre-taken photos exist for each node, then these photos will be packed into an inspection task of a drone for the drone to refer to when comparing or identifying wires. The advantages of this are as follows: on the one hand, it allows the drone to compare the known images with the real-time captured images during flight to identify whether there are newly added, missing, or damaged wires; on the other hand, it can better ensure the inspection coverage and prevent important nodes or branches from being missed.

[0094] S3. Control the drone to go to the starting point of the corresponding area, and then starting from the starting point, traverse each node and branch based on preset rules and record, and check the wires during the traversal.

[0095] At this stage, the system first navigates the drone to the starting node of the corresponding area, and then uses this node as the starting point to traverse the entire sub-network along the wire branches one by one based on the preset inspection order. The so-called "preset rules" can be an inspection route plan defined manually or by an algorithm. A typical way is to, in the top view, take the currently located node as the center of the circle and sequentially select the surrounding branches for detection in the clockwise direction; in other words, when the drone arrives at a node, it will sequentially switch from the "clockwise" angle around each connecting branch under the top view of this node to determine the next wire route to be inspected.

[0096] In the present application, another way can also be adopted. Similarly, in the local topology network, take the currently located node as the center of the circle and sequentially select the surrounding branches for detection in the clockwise direction.

[0097] Specifically, S3 includes sub-steps S31 - S37.

[0098] S31. The drone goes to the starting point of the corresponding area.

[0099] S32. The drone flies to the reference point corresponding to the utility pole based on image recognition.

[0100] Optionally, S32 includes the following sub-steps S321 - S324.

[0101] S321. When the drone arrives at the area corresponding to the starting point, obtain an image of the utility pole at the node.

[0102] S322. Based on image recognition, the drone determines the position of the utility pole in the image.

[0103] S323. The drone selects a position at a preset height above the utility pole as the reference point.

[0104] S324. The drone flies to the reference point.

[0105] The drone will first automatically fly to the starting position of the corresponding area according to the previously divided area information, and regard this position as the starting point of the subsequent inspection activities. The significance of this step is to establish a fixed reference position for the drone, enabling the entire inspection task to be carried out in a clear initial scenario. Then, the drone will use image recognition technology to go to the reference point corresponding to the utility pole. Here, "image recognition" refers to using machine vision algorithms to detect and locate targets within the field of view, such as utility poles, crossbeams, or other prominent landmarks, so as to provide accurate coordinate references for the drone's navigation. The reference point usually refers to a position at a certain preset height above the utility pole, ensuring that the drone neither interferes with the power grid facilities nor can obtain a stable and wide field of view for monitoring.

[0106] In the actual implementation process, once the drone reaches the area where the starting point is located, it will first take a top-down or side view of the area, and then use a special image recognition model to detect whether there is a utility pole in the image and its specific position. Assuming that the image recognition algorithm detects that the utility pole is located in a certain area of the picture, the drone will calculate its relative position and height relative to itself, and set the target height to a safe and observable height value, such as three to five meters higher than the top of the utility pole. Subsequently, the drone flies to this fixed point, that is, the reference point, according to the calculation result, and hovers within the safe area directly above the utility pole. Through such step arrangements, whether there are other obstacles in the environment or the shape of the utility pole itself is irregular, the flight path can be effectively calibrated relying on image recognition and reference point setting, ensuring that the drone can complete various shooting and detection tasks at an appropriate height.

[0107] S33. The drone creates a blank new topological network diagram and generates a node corresponding to the utility pole where it is located within the new topological network diagram.

[0108] S34. The drone adjusts to the preset attitude at the reference point and takes a picture at the node position to obtain an image of the node position.

[0109] When the drone arrives at the utility pole's reference point, it first creates a new, blank topological network map within its system and generates nodes corresponding to the pole it is currently viewing. A "topological network map" is a digital representation of utility poles and their connections, providing a more intuitive visual representation of the spatial structure and relationships of power lines. Since the drone initially lacks visibility into the specific locations of new and existing power lines, it can create this blank map to begin recording the nodes and related information it observes in real time, effectively laying a scalable foundation for subsequent inspection data. For example, if the drone arrives at a utility pole at location A, it will initially record "Node A" on the new topological map, which will serve as the starting point for subsequent line or node expansion. This allows the drone to automatically expand this new map even if additional poles or branches are discovered during subsequent inspections, avoiding confusion or redundancy in inspection data.

[0110] After the nodes are generated in the new topology, the drone will shoot at the reference point according to the pre-set target flight attitude. The so-called "flight attitude" is usually provided by sensors such as gyroscopes in real time. The attitude data includes the drone's pitch angle, yaw angle, and roll angle. These data will be used by the system to adjust the drone to a specific angle to ensure that the shooting direction or the camera's downward angle is consistent with the pre-set standard. For example, if a certain utility pole is best photographed from a 10-degree angle from directly above in order to clearly see the crossarm and the branching wires, then after reading the gyroscope data, the drone will automatically make fine adjustments and finally hover at the predetermined attitude before shooting.

[0111] 35. The drone uses the pre-recorded node image information as a priori conditions to identify the wires in the node location image.

[0112] Optionally, the S35 includes the following steps S351-S356:

[0113] S351. Obtain pre-recorded node image information; wherein, the pre-recorded node image includes a utility pole and multiple wires, and the pre-recorded node image is taken at a reference point corresponding to the utility pole with a preset posture.

[0114] S352. Obtain a node position image, wherein the node position image is captured at a reference point corresponding to the utility pole in a preset posture.

[0115] S353. Align the node position image with the pre-recorded node image;

[0116] S354. Wire detection is performed on the node position image to identify thin lines or arcs;

[0117] S355. Analyze the features of the detected line segments, and merge approximately continuous short line segments to obtain the wire contour. Among them, feature analysis includes length analysis, angle analysis, and continuity analysis.

[0118] S356. In the corresponding registered coordinate system, compare the starting points, ending points, directions, and lengths of the line segments in the two photos. When the similarity or matching degree meets the set threshold, it is determined to be the same wire. Among them, the matching degree includes endpoint distance, average offset, and angle difference.

[0119] If no matching object can be found or the matching degree is low, it is determined to be a newly added wire or a reduced wire. Among them, a newly added wire corresponds to a wire that exists in the node position image but does not exist in the pre - entered node image; a reduced wire corresponds to a wire that does not exist in the node position image but exists in the pre - entered node image.

[0120] In this process, first, it is necessary to obtain the pre - entered node image information in S351, that is, the historical images taken by the drone at the reference point corresponding to the telegraph pole with a preset attitude. Usually, these images are stored together with metadata such as the shooting angle and flight altitude for subsequent comparison. When the drone flies back to the same reference point, the newly taken node image can be compared with this pre - entered information. The reason it is called a "prior condition" is that this image records the initial wire and telegraph pole layout, which can provide a direct reference for judging whether the wires have increased or decreased. For example, if there were originally only two wire branches on a certain telegraph pole, but now there are three or only one in the newly taken image. That is to say, two of the three will still be in the predictable positions, or the remaining one will be in one of the positions of the two predicted wires. Then the system can more accurately determine the actual change in the number of wires based on the wire changes at this position.

[0121] In S352, the drone will obtain the current node position image in the same or approximate flight attitude and use it as a new data source for comparison. To achieve this, the drone will calibrate its attitude in advance according to the established shooting angle and flight altitude to ensure that the "node position image" has a high similarity with the pre - entered image in terms of viewing angle and resolution. The advantage of this is that it significantly reduces the error in the subsequent comparison of the two photos and improves the accuracy of wire contour matching at the same time.

[0122] Immediately enter S353 to perform image registration on the node position image and the pre-entered node image. The so-called "registration" refers to making precise corrections to two photos in terms of translation, rotation, or scale based on several detectable feature points in the images (such as obvious structures at the top of utility poles, recognizable markers in the background, etc.), so that they overlap as much as possible within the coordinate system. If the shooting positions and postures are almost exactly the same, the adjustment in this step usually only requires fine-tuning to achieve; if there are certain tilts or distance errors in the shooting, feature points can be detected and matched through algorithms such as SIFT and ORB, and homography transformation or other geometric correction methods can be applied to complete the alignment. For example, if dozens of matching feature points are found in two images, the system will infer how to scale, rotate, or translate one of the images based on these points so that the two images highly coincide at the pixel level.

[0123] When reaching S354, the drone needs to perform wire detection on the currently captured node position image to identify the slender lines or curves therein. In actual operation, either traditional image processing algorithms such as the Hough transform can be used, or a deep learning object detection model can be used to segment the wires. Regardless of which algorithm is used, the ultimate goal is to obtain the line segment or arc region that may represent the wire. To improve the accuracy of detection, the system often performs preliminary filtering or aggregation on the detection results and eliminates stray short line segments or noise points to retain a more continuous wire shape.

[0124] In S355, the system performs further feature analysis and merging on the detected wire segments. Feature analysis usually includes quantifying indicators such as the length, orientation (angle), and continuity of the line segments; when several line segments can be pieced together into a longer continuous wire in terms of angle and endpoint position, a merging operation will be performed to avoid recording multiple parts of what is originally one wire as multiple independent line segments. For example, in a top view, several short lines that are closely connected to each other may be detected. If the splicing angles between them are not very different and the adjacent endpoint distances are within a certain threshold range, the system will merge them into a wire contour. Through this aggregation, subsequent comparison and judgment can more accurately lock in the overall shape of the wire.

[0125] Finally, at S356, the system will perform a substantial line segment matching on the two registered images, including comparing key information such as the starting point, ending point, length, orientation, and angle of the line segments. If the matching degree between the two (such as endpoint offset, angle difference, average distance error, etc.) is within the preset threshold, it can be determined that this is the same wire; if no corresponding matching object can be found or the matching degree is very low, it is determined as a "newly added" or "reduced" wire. A newly added wire means a line segment that exists in the current image but is absent in the pre-entered image, which may be connected to a newly built utility pole or a new branch extended from an existing utility pole; a reduced wire indicates a line segment that originally existed but is no longer visible in the captured image, and is inferred to be interrupted or removed and lose its connection.

[0126] It should be noted that in this solution, the reason for emphasizing that the drone must be photographed at specific reference points and angles is to enable a one-to-one direct comparison between the captured node images and the pre-stored images of "the same utility pole, the same angle". Compared with fuzzy searching in multiple photos, this shooting method of "constant angle, fixed position" can minimize the influence of environmental variables and make the two images maintain higher consistency in terms of perspective, scale, shape distortion, etc. In this way, the matching algorithm can determine whether it is indeed the same utility pole and its attached wires in the picture, or whether there are new connections, at a lower cost and with higher accuracy. For example, if the shooting angle of the drone's pan-tilt is deviated too much, the height and inclination angle of the utility pole presented in the image will be very different from the pre-entered image, which is likely to cause the algorithm to be difficult to accurately identify or misclassify. By maintaining a consistent reference height, orientation, and shooting posture, the differences between the two images mainly come from the on-site equipment itself (such as the increase or decrease of new and old wires), rather than the distortion caused by the shooting posture. This not only reduces complex multi-image searching or large-scale matching, but also provides more pure reference information for subsequent determination of wire increase or decrease.

[0127] S36. The drone generates branches corresponding to the wires based on the node position images and generates nodes corresponding to the utility poles where it is located in the new topological network diagram.

[0128] In this step, the drone will generate corresponding wire branches in the new topological network diagram based on the images of the node positions captured and the results of comparative analysis, and improve or update the corresponding utility pole nodes. Since the comparison between the wire information in the old topological network and the current observed images has been completed in the previous stage, the system can easily identify the wires that have not changed, regard them as "known" and directly retain them; at the same time, it can quickly discover the wires that have actually been added or removed, and add or remove the relevant nodes and branches in the new topological diagram accordingly. Without such a comparison, the drone is like flying in an unfamiliar environment, and it is difficult to quickly determine which wires are original and which are newly emerged. All identifications and generations have to start from scratch, which is time-consuming and prone to omissions or confusion.

[0129] In fact, the shapes of wires are often very complex in the real environment. They may be in various states such as multiple parallel, bundled and coiled, sagging due to gravity or swaying with the wind. Without comparison with the old images, it is extremely easy for the system to have errors in identification. Once the originally known wires are filtered out through comparison, it becomes efficient and accurate to identify the parts that have actually changed, thus significantly reducing the redundant processes during re-mapping.

[0130] Meanwhile, the old topological network diagram also plays a crucial role in this link. It not only provides a clear flight range for the drone, enabling it to know the inspection boundary without relying on GPS positioning; more importantly, the old diagram provides a reliable route guidance framework for the generation of the new diagram. Since the overall skeletons of the two are basically the same, the drone only needs to make fine-tuning according to the locally detected differences on the originally set path, which not only avoids getting "lost" in the complex branches but also reduces possible repeated traversals. For example, if the new diagram completely abandons the old diagram and independently plans the flight, each branch may be inspected back and forth more than twice, seriously wasting battery life and time; relatively, if new nodes or branches are found in the new diagram, their change ranges are usually not large and will not cause a fundamental impact on the overall inspection. Thus, by comparing the old images and old topological data, it is possible to significantly save the costs of identification and flight, and effectively improve the perception accuracy of wire addition and removal in a changing environment.

[0131] S37. The drone determines the branch inspection order of the drone at this node based on the new topological network diagram, and inspects the wires corresponding to the branches according to the branch inspection order and inspection strategy.

[0132] Specifically, the inspection strategy includes:

[0133] The drone preferentially inspects the branches according to the branch inspection order set by the local topological network at this node. If there are still branches that have not been inspected after the inspection of the wires corresponding to the local topological network in the new topological network diagram, then inspect these branches.

[0134] The drone determines whether the node reached during the branch inspection is the corresponding node within the local topology network based on image recognition. If so, it determines whether all the branches corresponding to this node have been traversed. If not, it proceeds to the next branch in the branch inspection order. If so, it returns to the previous node.

[0135] If not, it calculates the distance between this node and the center point. If this distance is less than or equal to the distance to the edge, it determines that this node is a new node in this partition. If it is greater, it controls the drone to return to the previous node.

[0136] At this stage, the drone has identified each branch and the corresponding power pole nodes based on the new topology network diagram and needs to further plan its own inspection order to ensure comprehensive inspection coverage and avoid duplication. The so-called "branch inspection order" refers to how the drone makes sequential choices among numerous possible wire branches after reaching a certain node to achieve a balance between efficiency and coverage. Usually, the local topology network will first formulate a general branch order based on the original division and route. If there are still branches not included after new wires are built or existing wires are added or removed, the system will conduct a follow-up inspection of those remaining branches after completing the established inspection tasks.

[0137] In specific implementation, the drone first reads the information of each branch connected to the current node in the new topology network diagram and conducts sequential flight inspections along each branch according to the established order. When the drone reaches the next node along a branch, it determines whether the node it has reached is exactly the known node recorded in this local network diagram based on image recognition technology. If the match is successful, it checks whether there are still untraversed branches on this node. If so, it continues to the next branch in the same way. If all the branches of this node have been inspected, the drone returns to the previous node to avoid blindly shuttling back and forth in the network. If the image recognition result shows that the node reached by the drone is not in the existing records of the current local topology network, the system will also calculate the distance between this node and the center point to determine whether it should be regarded as a "newly added node". If the distance does not exceed the edge threshold, it will be added to the network diagram of this partition. Otherwise, the drone will determine that it is in other partitions or outside the effective inspection range and thus choose to return to the previous node.

[0138] Through such a mechanism, the drone can always rely on the existing information in the new topological network diagram for orderly navigation during the inspection process. It can not only make full use of the general framework of the inspection path provided by the old diagram but also make flexible adjustments when new nodes or branches are discovered, reducing unnecessary repeated flights and route conflicts. For example, if there are three branches recorded on node A and they have all been inspected, then when the drone flies along a branch that may not be recorded to an unknown node, it can judge in real time whether to include the node in the update of the current sub-network after arrival. If the distance is too large, it is very likely that the node belongs to other partitions (because the possibility of densely adding multiple utility poles within a small area in this partition is extremely low). In this case, it is selected to return directly to avoid wasting inspection time.

[0139] Optionally, S37 includes the following steps S371 - S373.

[0140] S371. Control the drone to obtain image information and obtain wire image information based on instance segmentation;

[0141] S372. Input it into a pre-trained deep learning model to identify the risk type;

[0142] S373. If a risk is identified, record the branch and mark the risk type, and take a photo of the risk location.

[0143] Among them, the training steps of the pre-trained deep learning model include S3721 - S3726.

[0144] S3721. Determine the risk types to be identified and define the recognizable features of each risk type;

[0145] S=3722. Obtain image samples containing recognizable features of risks and manually mark the recognizable features in the images using annotation tools;

[0146] S3723. Divide the image samples into a training set and a test set based on a pre-selected partitioning strategy and partitioning ratio. Among them, the training set is used to train the model, and the test set is used to verify and adjust the model. The pre-selected partitioning strategy is random partitioning, stratified partitioning, or using cross-validation;

[0147] S3724. Train a CNN model using the ResNet architecture based on the training set, and use the cross-entropy loss function during the training process;

[0148] S3725. Input the test set into the trained model, input the risk types identified by the trained model into the preset reward function for calculation to obtain a reward value. Among them, the output of the preset reward function is related to the difference between the risk types identified by the CNN model and the actual risk types. When the difference is larger, the reward value is lower; when the difference is smaller, the reward value is smaller.

[0149] S3726. Based on the reinforcement learning algorithm and the change of the reward value, adjust the parameters of the trained model to increase the reward value, so as to obtain new risk types, substitute them into the previous step and execute until the risk types converge to a stable range.

[0150] S4. Control the drone to generate a new partition topology network based on the recorded nodes and branches.

[0151] S5. After all drones complete the inspection tasks and return, obtain the nodes and branches recorded by the drones, determine the common intersection points generated by the drones in adjacent areas, summarize the partition networks into a total topology network, and delete the repeatedly generated branches and nodes.

[0152] After the drone completes the inspection of each node and branch in the partition, it will organize the captured images and the real-time recorded branch connection information, and then generate a new partition topology network diagram locally. The so-called "new partition topology network diagram" refers to constructing the nodes, routes and their interconnection relationships obtained by the current drone inspection of the partition in digital form, echoing or supplementing the original topology network. For example, if the drone discovers several new wires or new utility poles during the inspection process, it will immediately record these discoveries in the current partition network diagram to form an "incremental" update. The reason for not comprehensively photographing the entire area from a high altitude at the beginning and constructing or reconstructing the entire topology network at one time is that due to ground environment occlusion, cable sag and other complex factors, it is often difficult for aerial images to clearly capture every wire. Adopting the branch-level inspection method can more effectively master the actual distribution, thus avoiding errors or omissions to the greatest extent. At the same time, it also avoids the secondary work of first establishing a large and rough network structure and then conducting a detailed traversal, saving time and resources.

[0153] After all the drones have completed the inspection tasks for their respective areas and returned, the system will centrally summarize all the new area topology networks and captured images in the background. Since GPS positioning is not used (to prevent problems such as drift or unstable signals caused by precision errors), there may be overlaps in the junction areas of the area networks established by each drone during field operations and the captured images of utility poles. To accurately identify these overlapping or identical nodes as the same object, they need to be uniformly compared and de-duplicated during the summarization stage. That is to say, if different drones report the same utility pole or branch, the system will confirm whether they actually point to the same target through image recognition and node attribute comparison. Once they match, they will be merged into one node and duplicate records will be deleted. This can ensure that the final overall topology network obtained has no redundant nodes or wire branches, and at the same time, it can also merge the different discoveries reported by each drone (such as newly added wires or removed branches) into the overall topology map of the same power system. Finally, without the aid of GPS, by means of multi-drone collaborative area inspection, offline data transmission, and data fusion, an accurate, detailed, and up-to-date power system network structure can still be efficiently constructed and maintained.

[0154] An embodiment of the present application also discloses a power cable monitoring and analysis system, including a processor, and a program of the power cable monitoring and analysis method described in any one of the above is run in the processor.

[0155] An embodiment of the present application also discloses a storage medium, storing a program of the power cable monitoring and analysis method described in any one of the above.

[0156] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A method for monitoring and analyzing power cables, characterized in that, It includes the following steps: S1. Obtain the overall topological network of the power system and the number of drones, and partition the nodes of the overall topological network based on the number of drones to obtain multiple local topological networks. Here, the overall topological network includes nodes and branches, the nodes correspond to utility poles, and the branches correspond to wire branches; S2. Create inspection tasks and task data information based on the local topological networks and distribute them to each drone respectively. The task data information includes the node information and branch information corresponding to a certain local topological network, as well as the pre-entered node image information corresponding to each node. The edge of the local topological network is a node, and adjacent local topological networks are connected through shared nodes; S3. Control the drone to go to the starting point of the corresponding partition, and then start from the starting point, traverse each node and branch based on a preset rule and record, and check the wires during the traversal; S4. Control the drone to generate a new partition topological network based on the recorded nodes and branches; S5. After all drones complete the inspection tasks and return, obtain the nodes and branches recorded by the drones, determine the common intersections generated by drones in adjacent areas, summarize the partition networks into a total topological network, and delete the repeatedly generated branches and nodes; The S3 includes: S31. The drone goes to the starting point of the corresponding partition; S32. The drone flies to the reference point corresponding to the utility pole based on image recognition; S33. The drone creates a blank new topological network diagram and generates a node corresponding to the utility pole where it is located in the new topological network diagram; S34. The drone adjusts to a preset attitude at the reference point and takes a photo at the node position to obtain a node position image; S35. The drone uses the pre-entered node image information as a prior condition to identify the wires in the node position image; S36. The drone generates branches corresponding to the wires based on the node position image and generates nodes corresponding to the utility poles where it is located in the new topological network diagram; S37. The drone determines the branch inspection order of the drone at this node based on the new topological network diagram and checks the wires corresponding to the branches according to the branch inspection order and inspection strategy.

2. The power cable monitoring and analysis method according to claim 1, wherein The S1 includes the following steps: S11. Obtain the overall topological network of the power system and the number N of drones, and determine the branch number allocation range and node number allocation range of the local topological network based on the total number of nodes, total number of branches of the overall topological network, and the number of drones; S12. According to the branch number and distance constraints, make a preliminary judgment and screening for each node of the overall topological network, and select N candidate nodes as the possible centers of the local topological networks; S13. With each candidate center point as the core, absorb the unowned nodes into their respective sub-networks layer by layer outward or in the shortest path manner, where the absorption speed of each candidate center is the same; S14. When the number of nodes or branches in a certain sub-network reaches the upper limit, stop absorbing new nodes into this sub-network; S15. Allocate the remaining nodes without belonging to avoid the appearance of isolated nodes or the appearance of redundant sub-networks; S16. Inside each local topology network, check the path length from the central point to each edge node. If it exceeds or is lower than the preset threshold, then by means of boundary node adjustment, reassign some nodes to adjacent local topology networks so that the distance from the central point to the edge gradually returns to a reasonable range; S17. Re-evaluate the position of the central point so that the difference between the maximum node distance and the minimum node distance from the central point to the edge in the local topology network is less than the preset threshold; among them, the position of the central point serves as the starting point of the UAV flight in the partition; S18. Check the fulfillment of the constraints of the local topology network and perform step backtracking for adjustment.

3. The power cable monitoring and analysis method according to claim 2, characterized in that, The inspection strategy includes: The UAV preferentially checks the branches according to the branch inspection order set by the local topology network at this node. If there are still branches that have not been inspected after the wire inspection corresponding to the local topology network in the new topology network diagram is completed, then inspect this branch; The UAV judges whether the node reached by inspecting along the branch is the corresponding node in the local topology network based on image recognition. If so: judge whether all the branches corresponding to this node have been traversed. If not, go to the next branch according to the branch inspection order. If so, return to the previous node; If not: calculate the distance between this node and the central point. If this distance is less than or equal to the distance to the edge, then determine that this node is a new node in this partition. If it is greater, then control the UAV to return to the previous node.

4. The power cable monitoring and analysis method according to claim 3, wherein The S32 includes the following sub-steps: S321. When the UAV arrives at the area corresponding to the starting point, obtain the image of the telegraph pole at the node; S322. The UAV determines the position of the telegraph pole in the image based on image recognition; S323. The UAV selects the position at a preset height above the telegraph pole as the reference point; S324. The UAV flies to the reference point.

5. The power cable monitoring and analysis method according to claim 4, characterized in that, The S35 includes the following steps: S351. Obtain the pre-entered node image information; among them, the pre-entered node image contains a telegraph pole and multiple wires, and the pre-entered node image is taken at the reference point corresponding to the telegraph pole in a preset pose; S352. Obtain the node position image, where the node position image is taken at the reference point corresponding to the telegraph pole in a preset pose; S353. Register the node position image and the pre-entered node image; S354. Perform wire detection on the node position image to identify slender lines or arcs; S355. Perform feature analysis on the detected line segments and merge approximately continuous short line segments to obtain the wire contour, where the feature analysis includes length analysis, angle analysis, and continuity analysis; S356. In the corresponding registered coordinate system, compare the starting point, end point, direction, and length of the line segments in the two photos. When the similarity or matching degree meets the set threshold, it is determined to be the same wire; among them, the matching degree includes the endpoint distance, average offset, and angle difference; If no matching object can be found or the matching degree is low, then it is determined to be a new wire or a reduced wire; where the new wire corresponds to a wire that exists in the node position image but does not exist in the pre-entered node image; the reduced wire corresponds to a wire that does not exist in the node position image but exists in the pre-entered node image.

6. The power cable monitoring and analysis method according to claim 5, characterized in that, S37 includes the following steps: S371. Control the drone to obtain image information and obtain wire image information based on instance segmentation; S372. Input it into a pre-trained deep learning model to identify the risk type; S373. If a risk is identified, record the branch and mark the risk type, and take a photo of the risk location; Among them, the training steps of the pre-trained deep learning model include: S3721. Determine the risk types to be identified and define the recognizable features of each risk type; S3722. Obtain image samples containing risk recognizable features of various types, and manually mark the recognizable features in the images using annotation tools; S3723. Divide the image samples into a training set and a test set based on a pre-selected partitioning strategy and partitioning ratio. Among them, the training set is used to train the model, and the test set is used to verify and adjust the model. The pre-selected partitioning strategy is random partitioning, stratified partitioning, or using cross-validation; S3724. Train a CNN model using the ResNet architecture based on the training set, and use the cross-entropy loss function during the training process; S3725. Input the test set into the trained model, input the risk type recognized by the trained model into a preset reward function for calculation and obtain a reward value. Among them, the output of the preset reward function is related to the difference between the risk type recognized by the CNN model and the actual risk type. The greater the difference, the lower the reward value, and the smaller the difference, the smaller the reward value; S3726. Adjust the parameters of the trained model based on the reinforcement learning algorithm and the change of the reward value to increase the reward value, so as to obtain a new risk type, substitute it into the previous step and execute until the risk type converges to a stable range.

7. A power cable monitoring and analysis system, characterized in that, It includes a processor, and a program of the power cable monitoring and analysis method as described in any one of claims 1-6 runs in the processor.

8. A storage medium, characterized in that, Store a program of the power cable monitoring and analysis method as described in any one of claims 1-6.

Images