Substation inspection unmanned aerial vehicle path planning method based on fusion A*- grey wolf algorithm

Abstract

The invention discloses a transformer substation inspection unmanned aerial vehicle path planning method based on a fusion A *-grey wolf algorithm, and aims to solve the problem of unmanned aerial vehicle path planning in a transformer substation inspection scene, and the method fully considers coupling of task point traversal and obstacle avoidance, thereby ensuring that an inspection path allocated for unmanned aerial vehicle inspection accords with a real transformer substation scene. A transformer substation virtual environment model is established through a grid map of an A * algorithm, and path implementation of the algorithm is simplified; a population is initialized by improving Tent chaotic mapping, a grey wolf algorithm is improved by introducing an adaptive position updating strategy and an elite grey wolf optimization strategy, and an initial traversal path is generated; and refining through an A * algorithm dynamically weighted by an evaluation function to obtain a final path. Simulation results prove that the feasibility and effectiveness of the improved fusion algorithm in obstacle avoidance and path planning in a substation simulation scene, and the improved algorithm is more suitable for path planning in a complex environment with multiple task targets and multiple obstacles.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent substation intelligent inspection technology methods, and in particular to a substation inspection drone path planning method based on the fusion of A*-Grey Wolf algorithm. Background Art

[0002] Substation inspection is an important way to maintain the safe and stable operation of substations. In order to meet the increasing demand for efficient, real-time and accurate inspections of substations, drone inspections with the characteristics of low cost, high passability, strong timeliness and high precision are becoming a mainstream intelligent inspection method for substations. Although the scope of use of drone inspection technology in the field of substation inspections is expanding, the current drone inspections still rely on manual flight operations, and the overall flight path efficiency is low. There are many inspection points for drones in substations, and obstacles in substations have a great impact on the flight path. The path planning of drones has become an important scientific research content. How to find an optimal inspection path in a complex substation environment and complete the inspection task safely and efficiently in the shortest time is a key issue that needs to be solved urgently.

[0003] Drone path planning is a complex global optimization problem. Currently, drone path planning in power inspection environments is mainly used for transmission line inspections. However, unlike transmission line inspections, substation inspections are characterized by dense inspection points, complex environmental factors, and many obstacles. Traditional global path planning algorithms are mainly Dijkstra algorithms, A* algorithms, and various improved A* algorithms. The A* algorithm is the most effective direct solution algorithm for solving the shortest path in a static environment and is widely used in indoor and outdoor drone path planning. In the substation inspection scenario, the A* algorithm will fall into a complicated path search due to the dense distribution of inspection points and the complexity of obstacles, resulting in a long search time and too many path turns. Therefore, we need to improve the traditional algorithm or introduce intelligent algorithms to improve the algorithm's search efficiency and solution quality. Intelligent optimization algorithms are represented by particle swarm optimization, ant colony optimization, and Grey Wolf Optimization (GWO). They can quickly and randomly search in the target search space. Compared with traditional path planning algorithms, they can find the global optimal solution more efficiently and at a lower cost. They have been widely used in solving path planning problems, but they also have the problem of premature convergence and falling into the local optimal solution. However, when the existing research methods are used to solve the UAV path planning problem, the improved traditional A* algorithm adopted can solve the general path planning problem well. However, with the increase of task nodes and the complexity of obstacles in the substation, it is easy to fall into complicated iterations, resulting in low computational efficiency; different from the substation scenario, the points in the transmission scenario are scattered and the task points are linear. The local optimum that the algorithm falls into in the transmission environment may be the long-distance path selection between poles and towers. In the substation environment, the algorithm is easy to fail to take into account the path optimization of traversal points and obstacle avoidance, and fall into the local optimum; therefore, the intelligent algorithm suitable for the transmission scenario may not be applicable in the substation scenario, and cannot be well applied in the substation inspection scenario; therefore, it is necessary to design a substation inspection UAV path planning method based on the fusion of A*-Grey Wolf algorithm to solve the above problems. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a substation inspection drone path planning method based on the fusion of A*-grey wolf algorithm. The method solves the path planning and obstacle avoidance problems in the complex environment of the substation while considering the traversal of inspection points. The A* algorithm is introduced into the traversal process through a grid map, and the improved Tent chaotic map is used to initialize the population of the grey wolf algorithm. The adaptive position update strategy and the elite grey wolf optimization strategy are used to ensure the convergence speed of the algorithm and avoid falling into the local optimum, thereby achieving the optimization of the traversal path. The improved A* algorithm with dynamic weighted optimization of the evaluation function is used to further improve the path between nodes and complete the optimization of the obstacle avoidance path. The method aims to achieve efficient traversal of inspection points and automatic obstacle avoidance. Finally, experimental simulation is carried out in a virtual map of the substation to verify the feasibility and effectiveness of the A*-grey wolf algorithm.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a substation inspection drone path planning method based on the fusion of A*-Grey Wolf algorithm, comprising: S1, based on the actual substation task scenario, considering the traversal inspection path and obstacle avoidance, a path planning mathematical model is established with the shortest obstacle avoidance inspection path as the objective function; S2, transform the actual substation map into an abstract map that retains the main location information and characteristics: The virtual map is established based on the actual equipment locations in the substation, the actual equipment forms in the substation are fuzzily simplified, the equipment area is drawn according to the safety distance, and the grid map method is used to model the substation working environment; S3, introduce the A* algorithm into the traversal process through the grid map, use the improved A* algorithm with dynamic weighted optimization of the evaluation function to further improve the path between nodes and complete the optimization of the obstacle avoidance path; S4, using the improved Tent chaotic mapping strategy to initialize the population, introducing the adaptive position update strategy and elite gray wolf optimization strategy; S5, experimental simulation is carried out in the substation virtual map to verify the feasibility and effectiveness of the A*-Grey Wolf algorithm.

[0006] Preferably, in step S1, the method of establishing a path planning mathematical model taking into account traversal inspection paths and obstacle avoidance and taking the shortest obstacle avoidance inspection path as the objective function is: Drones in smart substations Inspection tasks are carried out for inspection. For substation inspection area The flight distance of the drone is related to the total flight range of the drone itself and the total inspection time of the substation. In order to minimize the total inspection time of the substation, the total flight distance of the drone is used as the objective function and the objective function is calculated. : ; in, and are the node coordinates at both ends of each flight path, is the number of nodes; Defining inspection tasks The weight coefficient is ; Since each inspection task has the same priority in the substation drone inspection problem, ;definition is the distance penalty factor of the inspection drone. The planned path objective function value needs to decrease monotonically with the increase of the number of iterations. The value of is less than 1, ; The total reward strategy for the drone to traverse all inspection points is: ; in, is the shortest obstacle avoidance path length between two inspection points of the UAV; To avoid obstacles, the shortest obstacle avoidance path strategy is adopted, and the safety factor is introduced to ensure the safe flight of the drone. The safety strategy expression is as follows: ; in, is the safety factor, is the coordinate within the obstacle. When the UAV path passes through the obstacle coordinate, it means that the path collides with the obstacle and the safety factor is reduced. In this case, the shortest obstacle avoidance path of the path is increased to bypass the obstacle and ensure the safety of the UAV flight.

[0007] Preferably, the substation virtual environment map model in step S2 is established as follows: The substation scene is fixed and static, but the huge space of the substation is not conducive to the fixed-point inspection of the drone during actual flight, so it is necessary to convert the actual map of the substation into an abstract map that retains the main location information and characteristics; the virtual map is established in combination with the actual equipment locations in the station. The substation is equipped with a main control building, high-voltage room, transformer, reactive compensation equipment and equipment area. Considering the complexity of the equipment form and lines, the actual equipment form in the station is fuzzy and simplified, and a block diagram of the equipment area is drawn according to the safe distance; considering the substation fence and no-fly zone, a virtual map of the substation inspection is drawn to obtain a 30×30 grid map.

[0008] Preferably, the grid map in step S3 introduces the A* algorithm into the traversal process, and the improved A* algorithm using the dynamic weighted optimization of the evaluation function improves the path between nodes in the following manner: Due to the flexibility of drones, eight-neighborhood search is used to search nodes in substation inspection. The node with the smallest path cost is selected as the next node through cost evaluation, and the nodes are searched one by one until the target inspection point is reached. The evaluation function of the A* algorithm is: ; in, Indicates the actual consumption cost from the starting point to the current node. Indicates the estimated cost from the current node to the target node. Represents the sum of the costs from the starting point to the current node and from the current node to the target node; Actual cost function Usually expressed as Euclidean distance: ; The commonly used expression of the heuristic function is Manhattan distance: ; in, is the horizontal coordinate of the target node, is the vertical coordinate of the target node, is the horizontal coordinate of the current node, is the ordinate of the current node; In path planning, the smaller the difference between the actual cost and the estimated cost between the target node and the current node, the better the path search effect; when the target node and the current node need to be planned over a long distance, the search requires more nodes and consumes a larger search space. The actual cost is much greater than the estimated cost, and the search efficiency is slowed down; when the distance between the target node and the current node is shortened, the search nodes and search space are also reduced accordingly, the actual cost is less than the estimated cost, and the search focus tends to search for the optimal path, but it is also easy to fall into the local optimum; therefore, in A dynamic weight coefficient is introduced before , dynamically change the weight of the estimated value according to the node location information to achieve the effect of balancing the actual value and the estimated value; in the substation inspection scenario, the estimated value is also affected by obstacles, and the distance and obstacle influence are set on the weight coefficient. The evaluation function of the improved A* algorithm is as follows: ; In the formula, Represents the Euclidean distance between the target node and the current node. Indicates the distance from the starting point to the current node; 0.8 is the initial weight coefficient, which means that when approaching the end point, the optimal path is given priority; is the number of grids in the matrix area with the target node and the current node as the diagonal; is the number of obstacle grids within the matrix area.

[0009] Specifically, the dynamic weighted optimization principle is to focus on the area where the target node is located in the initial stage of path search; after half of the search stage, the focus will be shifted to the optimization of the optimal path, and the search intensity of the path quality will be dynamically increased considering the number of obstacles, so as to achieve the effect of balancing search speed and search quality.

[0010] Preferably, in step S4, the improved Tent chaotic mapping strategy is used to initialize the population, and the adaptive position update strategy and the elite gray wolf optimization strategy are introduced in the following manner: The gray wolf algorithm is improved according to the position update formula, and an adaptive position update strategy is proposed to adaptively adjust the position of the gray wolf individual according to the fitness value of the gray wolf individual; the specific method is: The fitness value of the current gray wolf individual The average fitness value of the gray wolf population For comparison, if Lower than , then the original strategy is used to update the location; if Better than , then based on Wolf, Wolf and The wolves update their positions according to their fitness. During the update process, individuals with lower fitness in the gray wolf population will have a higher probability of learning other position information. The improved strategy equation is as follows: ; ; in, , and They are Wolf, Wolf and The fitness value of the wolf; when the fitness of the individual gray wolf is lower than the average fitness of the gray wolf population, in the process of position update Wolf, Wolf and The wolf makes the same contribution as before; if the fitness of a gray wolf individual is higher than the average fitness, then during the position update process, the gray wolf makes different degrees of contribution according to the size of its fitness.

[0011] Compared with the prior art, the beneficial effects of the present invention include: The present invention provides a substation inspection UAV path planning method based on the fusion of A*-Grey Wolf algorithm, which can adapt to the complex environment of the substation, meet the substation inspection needs, traverse the substation inspection points and take obstacle avoidance into consideration, and reasonably plan UAVs to conduct substation inspections, providing guidance for substation UAV inspection path planning in actual projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 It is the layout diagram of the substation equipment in the present invention; Figure 2 The substation simulation grid diagram of the present invention; Figure 3 A* algorithm search schematic diagram in the grid graph of the present invention; Figure 4 The social hierarchy pyramid of gray wolves in the present invention; Figure 5 The fusion algorithm process in the present invention; Figure 6 Schematic diagram of the task target point in the present invention; Figure 7 The present invention optimizes the sorting of task points based on the improved grey wolf algorithm; Figure 8 The task target point path planning diagram in the present invention; Fig. 9 The present invention integrates the iterative convergence graph of the A*-Grey Wolf algorithm; Fig.10 Comparison of algorithm running time for small-scale tasks in this invention; Fig.11 Comparison of algorithm running time under large-scale tasks in the present invention; Fig.12 Comparison of algorithm running time under different obstacle ratios in the present invention. DETAILED DESCRIPTION

[0013] A substation inspection drone path planning method based on the fusion of A*-grey wolf algorithm, comprising: S1, based on the actual substation task scenario, considering the traversal inspection path and obstacle avoidance, a path planning mathematical model is established with the shortest obstacle avoidance inspection path as the objective function; S2, transform the actual substation map into an abstract map that retains the main location information and characteristics: The virtual map is established based on the actual equipment locations in the substation, the actual equipment forms in the substation are fuzzily simplified, the equipment area is drawn according to the safety distance, and the grid map method is used to model the substation working environment; S3, introduce the A* algorithm into the traversal process through the grid map, use the improved A* algorithm with dynamic weighted optimization of the evaluation function to further improve the path between nodes and complete the optimization of the obstacle avoidance path; S4, using the improved Tent chaotic mapping strategy to initialize the population, introducing the adaptive position update strategy and elite gray wolf optimization strategy; S5, experimental simulation is carried out in the substation virtual map to verify the feasibility and effectiveness of the A*-Grey Wolf algorithm.

[0014] Preferably, in step S1, the method of establishing a path planning mathematical model taking into account traversal inspection paths and obstacle avoidance and taking the shortest obstacle avoidance inspection path as the objective function is: Drones in smart substations Inspection tasks are carried out for inspection. For substation inspection area The flight distance of the drone is related to the total flight range of the drone itself and the total inspection time of the substation. In order to minimize the total inspection time of the substation, the total flight distance of the drone is used as the objective function and the objective function is calculated. : ; in, and are the node coordinates at both ends of each flight path, is the number of nodes; Defining inspection tasks The weight coefficient is ; Since each inspection task has the same priority in the substation drone inspection problem, ;definition is the distance penalty factor of the inspection drone. The planned path objective function value needs to decrease monotonically with the increase of the number of iterations. The value of is less than 1, ; The total reward strategy for the drone to traverse all inspection points is: ; in, is the shortest obstacle avoidance path length between two inspection points of the UAV; To avoid obstacles, the shortest obstacle avoidance path strategy is adopted, and the safety factor is introduced to ensure the safe flight of the drone. The safety strategy expression is as follows: ; in, is the safety factor, is the coordinate within the obstacle. When the UAV path passes through the obstacle coordinate, it means that the path collides with the obstacle and the safety factor is reduced. In this case, the shortest obstacle avoidance path of the path is increased to bypass the obstacle and ensure the safety of the UAV flight.

[0015] Preferably, the substation virtual environment map model in step S2 is established as follows: In order to retain the characteristics of the actual substation to the greatest extent, the present invention establishes a virtual map in combination with the actual equipment locations in the station, fuzzily simplifies the actual equipment forms in the station, and uses the grid map method to model the substation working environment. The substation scene is fixed and static, but the huge space of the substation is not conducive to the fixed-point inspection of the drone during the actual flight of the drone, so it is necessary to convert the actual map of the substation into an abstract map that retains the main location information and characteristics. The virtual map is established in combination with the actual equipment locations in the station. The substation is distributed with a main control building, a high-voltage room, a transformer, reactive compensation equipment, and various equipment areas. Considering the complexity of the equipment form and the lines, the actual equipment form in the station is fuzzily simplified, and a block diagram of the equipment area is drawn according to the safe distance, as shown in the figure. Figure 1 shown.

[0016] Considering the substation fence and no-fly zone, a virtual substation inspection map is drawn to obtain a 30×30 grid map as shown in Figure 2 As shown; the black area is the boundary of the substation, the gray area represents the obstacle area and no-fly zone such as power equipment or fences, and the white area represents the area where passage is allowed; The coordinate range of the flight node is shown in the gray area, and the value range of the flight node is shown in the white area. The establishment of a virtual map can accurately locate the inspection task points and provide an environmental basis for subsequent path planning experimental simulations.

[0017] Preferably, the A* algorithm is a classic heuristic search algorithm, which is widely used in static map environments. It performs heuristic search based on search neighborhoods and cost functions under known environmental parameters, and the search is a four-neighborhood or eight-neighborhood search. The schematic diagram of the four-neighborhood and eight-neighborhood search represented by the grid graph method is as follows: Figure 3 shown.

[0018] The grid map in step S3 introduces the A* algorithm into the traversal process, and the improved A* algorithm with dynamic weighted optimization of the evaluation function is used to improve the path between nodes in the following way: Due to the flexibility of drones, eight-neighborhood search is used to search nodes in substation inspection. The node with the smallest path cost is selected as the next node through cost evaluation, and the nodes are searched one by one until the target inspection point is reached. The evaluation function of the A* algorithm is: ; in, Indicates the actual consumption cost from the starting point to the current node. Indicates the estimated cost from the current node to the target node. Represents the sum of the costs from the starting point to the current node and from the current node to the target node; Actual cost function Usually expressed as Euclidean distance: ; The commonly used expression of the heuristic function is Manhattan distance: ; in, is the horizontal coordinate of the target node, is the vertical coordinate of the target node, is the horizontal coordinate of the current node, is the ordinate of the current node; In path planning, the smaller the difference between the actual cost and the estimated cost between the target node and the current node, the better the path search effect; when the target node and the current node need to be planned over a long distance, the search requires more nodes and consumes a larger search space. The actual cost is much greater than the estimated cost, and the search efficiency is slowed down; when the distance between the target node and the current node is shortened, the search nodes and search space are also reduced accordingly, the actual cost is less than the estimated cost, and the search focus tends to search for the optimal path, but it is also easy to fall into the local optimum; therefore, in A dynamic weight coefficient is introduced before , dynamically change the weight of the estimated value according to the node location information to achieve the effect of balancing the actual value and the estimated value; in the substation inspection scenario, the estimated value is also affected by obstacles, and the distance and obstacle influence are set on the weight coefficient. The evaluation function of the improved A* algorithm is as follows: ; In the formula, Represents the Euclidean distance between the target node and the current node. Indicates the distance from the starting point to the current node; 0.8 is the initial weight coefficient, which means that when approaching the end point, the optimal path is given priority; is the number of grids in the matrix area with the target node and the current node as the diagonal; is the number of obstacle grids within the matrix area.

[0019] Specifically, the dynamic weighted optimization principle is to focus on the area where the target node is located in the initial stage of path search; after half of the search stage, the focus will be shifted to the optimization of the optimal path, and the search intensity of the path quality will be dynamically increased considering the number of obstacles, so as to achieve the effect of balancing search speed and search quality.

[0020] Furthermore, the GWO algorithm is initialized by generating an initial set of random candidate solutions; in each iteration, the best individual, the second best individual, and the third best individual in the population are called Wolf (leader), Wolf and wolves, and the rest of the individuals that make up the pack are called Wolf; The social hierarchy of pack wolves plays a vital role in the process of group hunting. The wolf will be Wolf, Wolf and The social hierarchy of a wolf pack is as follows: Figure 4 shown.

[0021] The group hunting behavior of wolves is divided into three stages: chasing, encircling and hunting. The mathematical model of wolves gradually approaching their prey is shown in the following formula: ; in, is the current iteration number, For prey in The position at the iteration, and The wolf is in Second and The position at the iteration. is a closed step, where and The definition of is as follows: ; ; in, and is the coefficient, is the distance control parameter, which decreases linearly from 2 to 0 as the number of iterations increases. and is a random number between [0,1].

[0022] During the hunting process, assuming that the three best wolves have a better understanding of the location of the prey, the other wolves follow the three leading wolves to surround and hunt the prey, and the position update is as follows: ; ; Since the GWO algorithm is initialized by randomly generating the initial solution, poor quality or poorly random solutions may appear, resulting in a slow convergence of the algorithm. The present invention uses the Tent chaotic mapping method to initialize GWO, which has better randomness than simple random generation and a simpler structure than other chaotic mappings, which can ensure the quality of the initial solution generation and the convergence speed of the algorithm; however, the Tent mapping has a small period and is prone to decay into a periodic sequence in the later stages of iteration. In order to avoid this situation and further improve the quality of the initial population, the improved Tent chaotic mapping is used for population initialization operations. The improved Tent chaotic mapping is shown in the following formula: ; in, Indicates the number of mappings, Indicates The function value of the sub-mapping. And introduce the random variable ,While increasing the controllable randomness of Tent chaos, it also ensures the regularity of mapping.

[0023] Preferably, in step S4, the improved Tent chaotic mapping strategy is used to initialize the population, and the adaptive position update strategy and the elite gray wolf optimization strategy are introduced in the following manner: The gray wolf algorithm is improved according to the position update formula, and an adaptive position update strategy is proposed to adaptively adjust the position of the gray wolf individual according to the fitness value of the gray wolf individual; the specific method is: The fitness value of the current gray wolf individual The average fitness value of the gray wolf population For comparison, if Lower than , then the original strategy is used to update the location; if Better than , then based on Wolf, Wolf and The wolves update their positions according to their fitness. During the update process, individuals with lower fitness in the gray wolf population will have a higher probability of learning other position information. The improved strategy equation is as follows: ; ; in, , and They are Wolf, Wolf and The fitness value of the wolf; when the fitness of the individual gray wolf is lower than the average fitness of the gray wolf population, in the process of position update Wolf, Wolf and The wolf makes the same contribution as before; if the fitness of a gray wolf individual is higher than the average fitness, then during the position update process, the gray wolf makes different degrees of contribution according to the size of its fitness.

[0024] Embodiment 2: Since all individuals in the gray wolf population are led by the best three gray wolves, when the best gray wolf individual falls into a local optimal solution, the entire gray wolf population may fall into the local optimal solution.

[0025] To solve this problem, the present invention proposes an elite individual update strategy. After each iteration, the top three gray wolves in the current gray wolf population are selected and recorded as new gray wolves. , and The three newly generated elite gray wolves are not obtained by comparison with the previous gray wolves, but are selected from the entire current gray wolf population. Using this elite individual update strategy can not only maintain population diversity, but also improve population search efficiency, greatly reduce dependence on the leadership of the first generation of gray wolves, and thus reduce the risk of falling into a local optimal solution.

[0026] The process of integrating A*-Grey Wolf algorithm is as follows Figure 5 shown.

[0027] The steps to improve the fusion algorithm are as follows: 1) Use the grid map method to model the environment, add obstacle areas, and determine the starting point and target point.

[0028] 2) Initialize the gray wolf parameters and introduce the Tent chaotic map to improve the randomness of population initialization and population diversity.

[0029] 3) Calculate the fitness values ​​of individual gray wolves and sort them, introduce an adaptive position update strategy to adaptively update the position information, and use the elite gray wolf optimization strategy to optimize the fitness value of the leader gray wolf.

[0030] 4) The location information and iteration times are updated. After searching, the gray wolf group compares the corresponding location information, iterates and filters, and obtains the traversal node path.

[0031] 5) The A* algorithm dynamically optimizes the paths between nodes with weights, generates the optimal path, and draws it in the virtual map.

[0032] 5. Use this algorithm to conduct experimental simulation in the substation virtual map to verify the feasibility and effectiveness of the A*-Grey Wolf algorithm.

[0033] Matlab R2023a software is used to simulate and verify the effectiveness of the improved fusion A*-Grey Wolf algorithm in the substation simulation environment. First, the starting point and target point are determined in the established 30×30 grid map. The starting point is (7,3), represented by a triangle symbol, and the task target point is represented by a circle symbol. The map has a total of 900 grid nodes, 225 obstacle grid nodes excluding the boundary box, and 675 passable grid nodes. The obstacle ratio is 25%. 23 inspection task points are set, and the point settings are as follows: Figure 6 shown.

[0034] according to Figure 6 The first stage of the hybrid algorithm uses the improved gray wolf optimization algorithm to optimize the sorting of global task points. The task point path optimization sorting in the simulation environment is as follows: Figure 7 shown.

[0035] Depend on Figure 7 It represents the optimized sorted initial path that traverses all task nodes calculated in the first stage of the algorithm, but it has not yet been applied to the grid environment, and the impact of obstacles and no-fly zones has not yet been considered. According to the initial sorting optimization of the task points, the second stage of the hybrid algorithm continues to refine the obstacle avoidance and the path between nodes, iteratively generates the optimal path, and the generated task target point path planning diagram is as follows: Figure 8 shown.

[0036] Depend on Figure 8 It can be seen that the algorithm is based on Figure 6 Point diagram and Figure 7 The initial task point path sorting was further used for multi-target path planning, and the UAV route was drawn in the grid map. The results show that the proposed algorithm has strong robustness in the substation simulation environment. It can select a suitable path for obstacle avoidance and target point traversal according to the obstacles and no-fly zones in the simulation map, and finally return to the take-off point. The algorithm convergence diagram of the shortest distance and average distance of the path calculation is shown in Figure 2. Fig. 9 shown.

[0037] In order to further verify the effectiveness of the algorithm proposed in this invention, the optimal paths and average path lengths of different algorithms are compared, and the results are shown in Table 1; and the calculation time of the algorithms is compared with different numbers of tasks, and the results are shown in Table 1. Fig.10 shown.

[0038] Table 1: Comparison of multiple run results of four algorithms

[0039] Combining Table 1 and Fig.10It can be seen intuitively from the data that the improved A* algorithm can achieve better computing results than the traditional A* algorithm. Although the Gray Wolf has better solution quality than the two A* algorithms, it requires a higher number of iterations and calculation time. The hybrid A*-Gray Wolf algorithm is superior to other algorithms in terms of solution quality and average number of iterations of path length. The fusion A*-Gray Wolf algorithm optimizes the optimal path length by 11.24% and 4.90% compared with the A* and Gray Wolf algorithms. In solving different task points, when the number of task points is small, the A* algorithm has a faster calculation speed, but as the number of task points increases, the A* algorithm falls into a complicated traversal calculation, and the calculation time increases significantly; when there are 25 task points, the calculation speed of the Gray Wolf is similar to that of the Improved A* algorithm. The calculation speed of the Fusion A* algorithm is slightly lower than that of the A* algorithm when the number of task points is low, but it is also better than the Gray Wolf algorithm. As the number of task points increases, the calculation speed advantage of the A*-Gray Wolf algorithm gradually increases, and it is better than the other algorithms in multi-task solving. The Fusion A*-Gray Wolf algorithm optimizes the speed of 25 task point path planning solution by 41.07% and 30.16% compared with the A* and Gray Wolf algorithms. In summary, in the small-scale task point optimization solution, the A*-Gray Wolf algorithm has a better calculation performance in solving such problems, but the other algorithms also have better performance in solving smaller-scale problems.

[0040] For larger-scale inspection tasks of substations, we further compared the performance of algorithms under large-scale problems, adjusted the grid scale, added task nodes, simulated the problem solving of algorithms under substations of different scales, and recorded the calculation time required for path planning of each algorithm under a larger scale environment. Fig.11 It can be seen that the increase in task nodes will lead to an increase in the algorithm running time. When there are fewer nodes, the running time of the A*-Grey Wolf algorithm is slightly less than the other algorithms, but still relatively close. After more than 150 nodes, the operation speed of the A*-Grey Wolf algorithm is significantly faster than the other three algorithms. When the number of nodes is 500, the running time is much less than the other algorithms, and it has a clear advantage in the algorithm's operation speed. This further shows that the proposed algorithm has better performance in solving large-scale problems.

[0041] In order to verify the applicability of the algorithm in different substations, the proportion of obstacles in the entire grid environment was changed, the path calculation of the algorithm in the complex obstacle environment of different substations was simulated, the number of task nodes was kept consistent, the obstacle proportion was changed, and the curve of the algorithm running time changing with the obstacle proportion (20%-40%) was tested as shown in the figure. Fig.12 shown.

[0042] Depend on Fig.12It can be seen that as the proportion of obstacles increases, the running speed of the algorithm gradually slows down. The running time increases rapidly at 20%-25%, and the A*-Grey Wolf algorithm slows down in the 25%-40% range, while the running time of the other algorithms still increases significantly as the proportion of obstacles increases. Whether it is the running time at a high obstacle ratio or the change trend at the end of the curve, the A*-Grey Wolf algorithm has a greater advantage, indicating that the proposed algorithm can be well applied in a variety of substation scenarios.

[0043] In summary, the fusion A*-Grey Wolf algorithm has good computing performance in simulating large-scale tasks of different substations and high obstacle ratio environments, and has obvious optimization effects in searching the shortest path length, average path length, average number of iterations and search time.

Claims

1. A substation inspection drone path planning method based on the fusion of A*-Grey Wolf algorithm, characterized in that: include: S1, based on the actual substation task scenario, considering the traversal inspection path and obstacle avoidance, a path planning mathematical model is established with the shortest obstacle avoidance inspection path as the objective function; S2, transform the actual substation map into an abstract map that retains the main location information and characteristics: The virtual map is established based on the actual equipment locations in the substation, the actual equipment forms in the substation are fuzzily simplified, the equipment area is drawn according to the safety distance, and the grid map method is used to model the substation working environment; S3, introduce the A* algorithm into the traversal process through the grid map, use the improved A* algorithm with dynamic weighted optimization of the evaluation function to further improve the path between nodes and complete the optimization of the obstacle avoidance path; S4, using the improved Tent chaotic mapping strategy to initialize the population, introducing the adaptive position update strategy and elite gray wolf optimization strategy; S5, experimental simulation is carried out in the substation virtual map to verify the feasibility and effectiveness of the A*-Grey Wolf algorithm.

2. According to claim 1, a substation inspection drone path planning method based on the fusion of A*-Grey Wolf algorithm is characterized in that: In step S1, the method of establishing a path planning mathematical model taking into account traversal inspection paths and obstacle avoidance and taking the shortest obstacle avoidance inspection path as the objective function is: Drones in smart substations Inspection tasks are carried out for inspection. For substation inspection area The flight distance of the drone is related to the total flight range of the drone itself and the total inspection time of the substation. In order to minimize the total inspection time of the substation, the total flight distance of the drone is used as the objective function and the objective function is calculated. : ； in, and are the node coordinates at both ends of each flight path, is the number of nodes; Defining inspection tasks The weight coefficient is ; Since each inspection task has the same priority in the substation drone inspection problem, ;definition is the distance penalty factor of the inspection drone. The planned path objective function value needs to decrease monotonically with the increase of the number of iterations. The value of is less than 1, ; The total reward strategy for the drone to traverse all inspection points is: ； in, is the shortest obstacle avoidance path length between two inspection points of the UAV; To avoid obstacles, the shortest obstacle avoidance path strategy is adopted, and the safety factor is introduced to ensure the safe flight of the drone. The safety strategy expression is as follows: ； in, is the safety factor, is the coordinate within the obstacle. When the UAV path passes through the obstacle coordinate, it means that the path collides with the obstacle and the safety factor is reduced. In this case, the shortest obstacle avoidance path of the path is increased to bypass the obstacle and ensure the safety of the UAV flight.

3. According to claim 2, a substation inspection drone path planning method based on the fusion of A*-Grey Wolf algorithm is characterized in that: The substation virtual environment map model of step S2 is established as follows: A virtual map is established based on the actual equipment locations in the substation. The substation is equipped with a main control building, a high-voltage room, a transformer, reactive compensation equipment and an equipment area. Considering the complexity of the equipment form and the lines, the actual equipment form in the station is fuzzily simplified, and a block diagram of the equipment area is drawn according to the safety distance. Considering the substation fence and the no-fly zone, a virtual inspection map of the substation is drawn to obtain a 30×30 grid map.

4. The method for substation inspection drone path planning based on the fusion of A*-Grey Wolf algorithm according to claim 3 is characterized in that: The grid map in step S3 introduces the A* algorithm into the traversal process, and the improved A* algorithm using the dynamic weighted optimization of the evaluation function is used to improve the path between nodes in the following manner: Due to the flexibility of drones, eight-neighborhood search is used to search nodes in substation inspection. The node with the smallest path cost is selected as the next node through cost evaluation, and the nodes are searched one by one until the target inspection point is reached. The evaluation function of the A* algorithm is: ； in, Indicates the actual consumption cost from the starting point to the current node. Indicates the estimated cost from the current node to the target node. Represents the sum of the costs from the starting point to the current node and from the current node to the target node; Actual cost function Usually expressed as Euclidean distance: ； The commonly used expression of the heuristic function is Manhattan distance: ； in, is the horizontal coordinate of the target node, is the vertical coordinate of the target node, is the horizontal coordinate of the current node, is the ordinate of the current node; exist A dynamic weight coefficient is introduced before , dynamically change the weight of the estimated value according to the node location information to achieve the effect of balancing the actual value and the estimated value; in the substation inspection scenario, the estimated value is also affected by obstacles, and the distance and obstacle influence are set on the weight coefficient. The evaluation function of the improved A* algorithm is as follows: ； In the formula, Represents the Euclidean distance between the target node and the current node. Indicates the distance from the starting point to the current node; 0.8 is the initial weight coefficient, which means that when approaching the end point, the optimal path is given priority; is the number of grids in the matrix area with the target node and the current node as the diagonal; is the number of obstacle grids within the matrix area.

5. The method for substation inspection drone path planning based on the fusion of A*-Grey Wolf algorithm according to claim 4 is characterized in that: In step S4, the improved Tent chaos mapping strategy is used to initialize the population, and the adaptive position update strategy and the elite gray wolf optimization strategy are introduced as follows: The gray wolf algorithm is improved according to the position update formula, and an adaptive position update strategy is proposed to adaptively adjust the position of the gray wolf individual according to the fitness value of the gray wolf individual; the specific method is: The fitness value of the current gray wolf individual The average fitness value of the gray wolf population For comparison, if Lower than , then the original strategy is used to update the location; if Better than , then based on Wolf, Wolf and The wolves update their positions according to their fitness. During the update process, individuals with lower fitness in the gray wolf population will have a higher probability of learning other position information. The improved strategy equation is as follows: ； ； in, , and They are Wolf, Wolf and The fitness value of the wolf; when the fitness of the individual gray wolf is lower than the average fitness of the gray wolf population, in the process of position update Wolf, Wolf and The wolf makes the same contribution as before; if the fitness of a gray wolf individual is higher than the average fitness, then during the position update process, the gray wolf makes different degrees of contribution according to the size of its fitness.

Images