Unmanned aerial vehicle-boat cooperative offshore wind plant autonomous inspection path planning method

Abstract

The invention discloses an unmanned aerial vehicle-boat cooperative offshore wind plant autonomous inspection path planning method, and belongs to the technical field of offshore wind power operation and maintenance. Dividing a to-be-inspected area formed by the determined to-be-inspected point set into a plurality of inspection sub-areas; unmanned aerial vehicle task allocation is carried out on the divided inspection subareas; then, an improved ant colony algorithm ACO is adopted to plan the flight path of the unmanned aerial vehicle in the inspection sub-regions and the navigation path of the unmanned ship between the inspection sub-regions; then, the partition scheme with the shortest time in the feasible partition schemes is selected; and finally, coping with the obstacle at the parking point of the unmanned ship and planning an obstacle avoidance path under an actual condition. According to the method, the bottleneck of insufficient endurance of single-platform unmanned aerial vehicle inspection is overcome, the inspection operation range is expanded, and the quality and efficiency of path planning are remarkably improved. And an effective path planning method is provided for unmanned aerial vehicle-boat cooperative autonomous inspection of the offshore wind plant.

Description

Technical Field

[0001] The present invention relates to offshore wind power operation and maintenance technology, and in particular to a method for planning autonomous inspection paths for offshore wind farms in collaboration with drones and boats. Background Art

[0002] With the growing global demand for clean energy, offshore wind power has rapidly developed as a key form of renewable energy. Offshore wind farms are typically massive, with numerous and widely distributed wind turbines. Regular inspection and maintenance are crucial to ensuring the safe and stable operation of wind turbines. Traditional inspection methods rely primarily on manual climbing, manned helicopters, or close-up observations from large vessels. These methods are inefficient and costly, and carry significant operational risks, especially in adverse sea conditions.

[0003] In recent years, drones, due to their high flexibility and relatively low cost, have been initially used in the inspection of key parts such as wind turbine blades, towers, and nacelles. However, the limited endurance of drones (especially electric multi-rotor drones) makes it difficult to meet the needs of comprehensive coverage and continuous inspections of large offshore wind farms. The coverage area of ​​a single flight is small, and frequent returns to recharge or replace batteries are required, which greatly reduces operational efficiency. In addition, the environment of offshore wind farms is complex, and the wind turbines themselves and other offshore facilities may pose obstacles to drone flights and unmanned boat navigation. Summary of the Invention

[0004] Purpose of the invention: In order to overcome the above limitations, a method for autonomous inspection path planning of offshore wind farms in collaboration with drones and boats is provided. The unmanned boat can carry drones for transfer within a vast wind farm and provide charging services for drones during navigation or mooring, thereby expanding the drone's operating range and continuous working capacity. First, based on the drone's endurance and inspection sensor characteristics, the wind farm area to be inspected is divided into several minimum inspection units and feasible inspection sub-areas; then, the unmanned boat acts as a mobile platform to be responsible for transporting drones between sub-areas and charging them during navigation or mooring; the drone cluster is responsible for performing specific inspection tasks within the sub-area. An improved ant colony algorithm is used to plan the flight path of the drone within the sub-area and the navigation path of the unmanned boat between the sub-areas, respectively, and local path adjustments under fixed obstacles in the wind farm (such as other wind turbine foundations and no-fly zones) are taken into account.

[0005] The above purpose is achieved through the following technical solutions:

[0006] A method for autonomous inspection path planning of offshore wind farms using UAV-boat collaboration is based on the following known quantities:

[0007] Parameters of the area to be inspected: length of the area to be inspected X max , Width of the area to be inspected Ymax ;

[0008] Rotor drone parameters: number of rotor drones m, rotor drone uniform flight speed v_a, rotor drone full charge E, rotor drone uniform charging speed β + , rotor UAV uniform discharge speed β - , Rotor UAV inspection flight altitude h, Optical sensor field of view angle

[0009] Unmanned boat parameters: Unmanned boat speed v_s;

[0010] Ant colony algorithm parameters: number of ants R, pheromone importance factor α, heuristic function importance factor β, pheromone attenuation coefficient ρ, total pheromone release Q, maximum number of iterations N_max;

[0011] The method comprises the following steps:

[0012] Step 1: Determine the set of points to be inspected;

[0013] Step 2: Divide the inspection area formed by the set of inspection points determined in step 1 into several inspection sub-areas;

[0014] Step 3: For the Kth inspection sub-area partitioning scheme (K=1, 2, ..., N), assign drone tasks to the divided inspection sub-areas;

[0015] Step 4: Use the improved ant colony algorithm ACO to plan the flight path of the UAV within the inspection sub-area and the navigation path of the UAV between the inspection sub-areas respectively;

[0016] Step 5: Select the partitioning scheme with the shortest time among the feasible partitioning schemes;

[0017] Step 6: Respond to obstacles at the unmanned boat berth and plan the obstacle avoidance path in actual situations.

[0018] Furthermore, step 1 specifically includes:

[0019] Define the square sea surface area that can be effectively covered by the optical sensor carried by the rotor UAV at the inspection flight altitude as the "minimum inspection unit", and define the projection point of the center of the minimum inspection unit at the flight altitude h as the inspection task point of the rotor UAV. Assuming that the wind farm area to be inspected is a rectangular area, the rotor UAV flies at a fixed altitude h, and the optical sensor carried by it is vertically downward, with a field of view angle of The side length d of the minimum inspection unit is calculated by the following formula:

[0020]

[0021] Define the total number of minimum inspection units as M, Indicates the minimum number of inspection units in the length direction. Indicates the minimum number of inspection units in the width direction, then:

[0022]

[0023] The projection points of the center points of these inspection minimum units at the flight height h constitute the UAV inspection task point set of the wind farm. Let the center coordinates of the jth inspection minimum unit be p j =(x j ,y j ,0),j=1,2,…,M, thus obtaining the task point set P that the rotor UAV cluster needs to inspect j =(x j ,y j ,h),j=1,2,…,M。

[0024] Furthermore, step 2 specifically divides each sub-area into the same size, first considering the minimum number of inspection units in the length and width directions of the area to be inspected. and Factor, assuming There are a total of factors of except 1, namely S1, S2, ..., S a ; There are b factors of excluding 1, namely f1, f2, ..., f b By permuting and combining these factors, theoretically, N=a×b different sub-region size specifications can be generated, where the sub-region corresponding to the Kth partitioning scheme contains s w ×f r The minimum inspection unit is w=1,2,…,a; r=1,2,…,b.

[0025] Furthermore, the specific method of step 3 is as follows:

[0026] For the Kth partitioning scheme, K = 1, 2, ..., N, the UAV tasks are assigned to the divided sub-areas. The rotorcraft UAV has endurance constraints and must meet the following conditions:

[0027]

[0028] D≤D max

[0029] Among them, E is the total power of the rotor UAV, v a is the uniform flight speed of the rotor UAV, β - is the uniform discharge speed of the rotor UAV, D maxis the maximum distance that a rotary-wing UAV can fly with full power, and D is the maximum flight path length of a single rotary-wing UAV performing a mission in a sub-area;

[0030] For a certain inspection sub-area, multiple drones conduct coordinated inspections. In order to make the time for each drone to perform tasks in the sub-area similar, thereby reducing the waiting time of the unmanned boat, the task allocation is based on the task point and the sub-area center, that is, the unmanned boat berthing point C. k The angle of the projection of the connecting line on the horizontal plane is set as the center point of the sub-area C k The coordinates of (x C ,y c ,0), the coordinates of a patrol task point A are (x A ,y A ,h); calculate the angle θ formed by the projection A′C of the line segment AC on the water surface and the counterclockwise rotation of the positive direction of the X-axis; if 2(l-1)π / m≤θ<2lπ / m, l=1,2,…,m, then the inspection task point is assigned to the lth UAV.

[0031] Furthermore, step 4 specifically includes the following sub-steps:

[0032] 4.1. UAV inspection path planning:

[0033] For each UAV u, the improved ant colony algorithm is used to plan its flight path at the inspection point set it is assigned to. The path starts from the UAV berthing point C k It takes off vertically to a height of h, traverses all the inspection points assigned to it, and finally returns to the unmanned boat berthing point C. k ;

[0034] make 1 means that UAV u in the Kth sub-area flies from task point i to task point j, otherwise it is 0, N ku is the number of task points that UAV u needs to inspect in sub-area K. The constraints include:

[0035]

[0036] Ensure that each inspection task point is only visited once by the corresponding drone;

[0037]

[0038] Make sure the drone is moving from the center of the sub-area C k take off and eventually return;

[0039] 4.2. Partition feasibility verification:

[0040] Calculate the length D of each planned drone path. If the path lengths D of all drones in the sub-area satisfy D≤D max, then the current sub-area division and the task allocation of the UAV cluster are feasible, traverse and verify all partitions and remove infeasible partitions;

[0041] 4.3. Unmanned Boat Path Planning:

[0042] If the current zoning scheme is feasible, all sub-area centers to be inspected, i.e., the berthing points C of the unmanned boats, will be k The improved ant colony algorithm is used to plan the shortest navigation path for the unmanned boat to start from the first anchorage point, visit all sub-area anchorage points once and finally return to the starting point.

[0043] 4.4 Calculate total task completion time

[0044] For the currently feasible sub-area division schemes and the corresponding UAV-boat collaborative path planning results, the total completion time t of the inspection task in the entire wind farm area under each division scheme is calculated. K (K=1,2,3,…,N), the formula is as follows:

[0045]

[0046] Δe=tβ -

[0047] Where n represents the total number of sub-areas divided in the Kth partitioning case; t represents the time required for the rotor UAV cluster to take off and land on the UAV in a sub-area (the flight time in each sub-area is the same); t i,i+1 represents the travel time of the unmanned boat between two adjacent sub-areas; t n,1 represents the time it takes for the UAV to return from the center of the last sub-area to the center of the first sub-area; Δe represents the actual maximum power consumption of the UAV; e1 represents the minimum power required for the UAV to take off; b + Indicates the charging speed of the drone; β - represents the UAV discharge speed, β + Indicates the charging speed of the drone;

[0048] The improved ant colony algorithm described in step 4.1 and step 4.3 is as follows:

[0049] 1. Initial pheromone concentration improvement, the formula is as follows:

[0050]

[0051] in, is the distance between the starting node and the target node, and are the distances between the starting node and the current node, and between the current node and the node to be selected;

[0052] 2. Elite strategy: Additional pheromones are added to the optimal path of the current iteration, and the pheromone update formula is changed to the following:

[0053]

[0054] Among them, e is the elite coefficient, Is the elite ant on the path (L i ,L j ) increases the amount of pheromones released, Node L at the current iteration i To node L j The pheromone concentration, Node L for the next iteration i To node L j The pheromone concentration, For the kth ant at node L i To node L j The pheromone released on the path, ρ is the pheromone attenuation coefficient, which represents the rate of pheromone volatilization on the path;

[0055] 3. Perform 2-opt optimization on the current optimal solution of each iteration to solve the problem of intersection of generated paths. If there are two local paths in the current optimal solution path of a certain iteration, they are: L i →L i +1 and L j →L j +1 if the exchange creates a new path L i →L j and the new path L i +1→L j +1, and meets: That is, if the length of the new path is shorter, it is replaced.

[0056] Furthermore, step 6 specifically includes the following sub-steps:

[0057] 6.1. What to do if there are obstacles at the unmanned boat anchorage:

[0058] If the unmanned boat is located at the planned berth point C in a sub-area under the optimal solution k If there is an obstacle at the stop point, then this sub-area division scheme is directly abandoned, and the second best division scheme is found among the feasible sub-area division schemes when there is no obstacle. If there is still an obstacle at the stop point of this division scheme, then the third best division scheme is searched, and so on, until a sub-area division scheme that meets the requirements is found;

[0059] If the unmanned boat is at the planned berthing point C in a certain sub-area k If there is no obstacle, go to step 6.2;

[0060] 6.2. Unmanned Boat Obstacle Avoidance Path Planning:

[0061] For the sub-area division scheme found, based on the overall path of the unmanned boat planned without obstacles, an improved ant colony algorithm with obstacle avoidance function is used to plan the overall obstacle avoidance path, as follows:

[0062] The obstacle repulsion term is introduced into the heuristic function:

[0063]

[0064] in, is node L j To the target node L g The Euclidean distance, λ is the weight coefficient, F r is the obstacle repulsion term, g is the number of obstacles within the field of view, d k is the distance from node j to the kth obstacle. The closer the obstacle is, the greater the repulsion;

[0065] Finally, the sub-area division scheme determined in step 6.1 and the final UAV and boat paths are selected as the optimal collaborative inspection path planning results in actual situations.

[0066] Beneficial effects:

[0067] (1) Aiming at the problem of limited endurance of drones in offshore wind farm inspections, the present invention proposes a drone-boat collaborative strategy based on sub-area division, which overcomes the bottleneck of insufficient endurance of single-platform drone inspections and expands the scope of inspection operations.

[0068] (2) This paper addresses the shortcomings of the traditional ant colony algorithm in terms of convergence speed and optimization ability, and proposes three key optimization methods to improve the ant colony algorithm: first, the initial pheromone concentration related to node distance is improved to accelerate the initial search process of the algorithm; second, the introduction of an elite strategy to add additional pheromones to the optimal path of the current iteration, accelerating the algorithm's convergence to a high-quality solution while maintaining sufficient diversity to avoid prematurely falling into a local optimal solution. Third, the 2-opt local optimization algorithm is combined to eliminate path intersections, significantly improving the quality and efficiency of path planning.

[0069] (3) Aiming at the problem of obstacles in the complex environment of offshore wind farms, the present invention proposes a local obstacle avoidance path planning mechanism based on heuristic function correction. By considering the modified heuristic function of the obstacle repulsion term, the algorithm's adaptability to fixed obstacles at sea is improved, and the algorithm's applicability in practical environments is enhanced.

[0070] (4) In order to solve the problem of insufficient adaptability of traditional path planning algorithms in cross-platform collaborative scenarios, the present invention proposes a complete systematic solution from inspection point determination, sub-area division to multi-platform collaborative path planning, providing an effective path planning method for the collaborative autonomous inspection of offshore wind farms by drones and boats. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a process of a method for planning a path for autonomous inspection of an offshore wind farm by coordinating a UAV and a boat, as described in the present invention.

[0072] Figure 2 This is an ideal obstacle-free simulation result diagram of the method of the present invention.

[0073] Figure 3 This is a diagram of the actual simulation result with obstacles using the method of the present invention.

[0074] Figure 4 This is a simulation diagram comparing the convergence curves of the improved ant colony algorithm of the method of the present invention and the traditional ant colony algorithm. DETAILED DESCRIPTION

[0075] See also Figure 1-Figure 3 As shown, the present invention provides a method for planning a path for autonomous inspection of an offshore wind farm by cooperating with a UAV and a boat, based on the known quantities:

[0076] Parameters of the area to be inspected: length of the area to be inspected X max , Width of the area to be inspected Y max ;

[0077] Rotor drone parameters: number of rotor drones m, rotor drone uniform flight speed v_a, rotor drone full charge E, rotor drone uniform charging speed β + , rotor UAV uniform discharge speed β - , rotor UAV inspection flight altitudej, optical sensor field of view angle

[0078] Unmanned boat parameters: Unmanned boat speed v_s;

[0079] Ant colony algorithm parameters: number of ants R, pheromone importance factor α, heuristic function importance factor β, pheromone attenuation coefficient ρ, total pheromone release Q, maximum number of iterations N_max.

[0080] Step 1: Determine the set of points to be inspected

[0081] When determining the point set for inspection, the present invention defines the square sea surface area that can be effectively covered by the optical sensor onboard the rotary-wing UAV at the inspection flight altitude as the "minimum inspection unit." The projection point of the center of the minimum inspection unit at the flight altitude h is defined as the inspection task point of the rotary-wing UAV.

[0082] Assuming that the wind farm area to be inspected is a rectangular area, the rotor UAV flies at a fixed height h, and the optical sensor it carries is pointed vertically downward, with a field of view of The side length d of the minimum inspection unit can be calculated by the following formula:

[0083]

[0084] Define the total number of minimum inspection units as M, Indicates the minimum number of inspection units in the length direction. Indicates the minimum number of inspection units in the width direction, then:

[0085]

[0086]

[0087] The projection points of the center points of these inspection minimum units at the flight height h constitute the UAV inspection task point set P of the wind farm. Let the center coordinates of these inspection minimum units be p j =(x j ,y j ,0),j=1,2,…,M, thus obtaining the task point set P that the rotor UAV cluster needs to inspect j =(x j ,y j ,h),j=1,2,…,M。

[0088] This inspection task point set determination method can transform the complex wind farm inspection area into a quantifiable and executable drone task point set, thereby achieving efficient and full-coverage inspection operations.

[0089] Step 2: Inspection sub-area division

[0090] Due to the power constraints of the rotor UAV, it is impossible to complete the inspection of the entire wind farm area in one flight, so sub-area division is necessary. The present invention divides each sub-area into the same size to simplify planning and scheduling.

[0091] Consider the minimum number of inspection units in the length and width directions of the area to be inspected and Assume that There are a total of a factors (except 1), namely S1, S2, ..., S a . There are b factors (except 1), namely f1, f2, ..., f bBy permuting and combining these factors, theoretically, N=a×b different sub-region size specifications can be generated, where the sub-region corresponding to the Kth partitioning situation contains s w ×f r A minimum inspection unit (w=1,2,…,a; r=1,2,…,b).

[0092] This sub-area division method optimizes inspection efficiency and task balance through a mathematical sub-area division method.

[0093] Step 3: Assign tasks to the corresponding drones for the task points in the sub-areas of the Kth partition

[0094] For the Kth partitioning scheme (K = 1, 2, ..., N), the drone tasks are assigned to the divided sub-areas. Rotary wing drones have endurance constraints and must meet the following conditions:

[0095]

[0096] D≤D max

[0097] Among them, E is the total power of the rotor UAV, v a is the uniform flight speed of the rotor UAV, β - is the power consumption rate of the rotor UAV (assuming uniform discharge), D max is the maximum distance that a rotorcraft UAV can fly with a full charge, and D is the maximum flight path length of a single rotorcraft UAV performing a mission in a sub-area.

[0098] For a certain inspection sub-area, multiple drones conduct coordinated inspections. To make the time for each drone to perform tasks in the sub-area similar, thereby reducing the waiting time of the drone, the task allocation is based on the task point and the sub-area center (drone berthing point C). k ) is projected on the horizontal plane. Let the center point of the sub-area be C k The coordinates of (x C ,y C ,0), the coordinates of a patrol task point A are (x A ,y A ,h). Calculate the projection A of line segment AC on the ground ′ The angle θ formed by C and the positive direction of the X axis in a counterclockwise rotation. If 2(l-1)π / m≤θ<2lπ / m, l=1,2,…,m. Then the inspection task point is assigned to the lth drone.

[0099] This task allocation method takes into account the balanced distribution of tasks and efficient collaborative inspections within sub-areas under the constraints of drone endurance.

[0100] Step 4: Plan the path for drone clusters and unmanned boats without obstacles and verify the feasibility of the zoning

[0101] 4.1 UAV Inspection Path Planning

[0102] For each UAV u, the flight path is planned using the improved ant colony algorithm (see step 4.5 for details). The path starts from the UAV mooring point C k It takes off vertically to a height of h, traverses all the inspection points assigned to it, and finally returns to the unmanned boat berthing point C. k .

[0103] make 1 indicates that UAV u in the Kth sub-area flies from task point i to task point j, otherwise it is 0. ku is the number of task points that UAV u needs to inspect in sub-area K. The constraints include:

[0104]

[0105] Ensure that each inspection task point is only visited once by the corresponding drone.

[0106]

[0107] Make sure the drone is moving from the center of the sub-area C k Take off and eventually return.

[0108] 4.2 Partition feasibility verification

[0109] Calculate the length D of each planned drone path. If the D of all drones in the sub-area satisfies D≤D max , then the current sub-area division and the task allocation of the drone cluster are feasible. Traverse and verify all partitions and remove infeasible partitions.

[0110] 4.3 Unmanned Boat Path Planning

[0111] If the current zoning scheme is feasible, all sub-area centers to be inspected (i.e., the berthing point C of the unmanned boat) k ) are considered as the path points that the UAV needs to traverse. The improved ant colony algorithm (see step 4.5 for details) is used to plan the shortest navigation path for the UAV starting from the first anchorage point, visiting all anchorage points in the sub-area once, and finally returning to the starting point.

[0112] 4.4 Calculate total task completion time

[0113] For the currently feasible sub-area division schemes and the corresponding UAV-boat collaborative path planning results, the total completion time t of the inspection task in the entire wind farm area under each division scheme is calculated. K(K=1,2,3,…,N), the formula is as follows:

[0114]

[0115] Where n represents the total number of sub-areas divided in the Kth partitioning case; t represents the time required for the rotor UAV cluster to take off and land on the UAV in a sub-area (the flight time in each sub-area is the same); t i,i+1 represents the travel time of the unmanned boat between two adjacent sub-areas; t n,1 represents the time it takes for the UAV to return from the center of the last sub-area to the center of the first sub-area; Δe represents the actual maximum power consumption of the UAV; e1 represents the minimum power required for the UAV to take off; b + Indicates the charging speed of the drone; β - Indicates the discharge speed of the drone.

[0116] 4.5 Improved Ant Colony Algorithm

[0117] The present invention adopts an improved ant colony algorithm (ACO) to plan the flight path of the UAV within the sub-area and the navigation path of the unmanned boat between the sub-areas respectively.

[0118] Ant colony algorithm works like this: ants release pheromones while foraging. Other ants can sense these pheromones and tend to choose paths with high pheromone concentrations. The shorter the path, the more ants travel back and forth per unit time, and the faster the pheromone accumulates. This creates a positive feedback loop, ultimately guiding the ant colony to find the optimal path.

[0119] Traditional ant colony algorithm steps:

[0120] 1. Path node selection probability: Assuming the number of ants is R, and the rth ant is currently at node i, the probability that it chooses to reach node j next is:

[0121]

[0122] Among them allowed r is the set of points that the rth ant can reach from its current node, τ ij is the pheromone on the path between two nodes, α is the pheromone importance factor, the larger α is, the greater the probability that the ant will choose the path with high pheromone concentration; η ij is the heuristic function from node i to node j, d ij is the Euclidean distance from node i to node j, β is the importance factor of the heuristic function, and the larger β is, the greater the probability that the ant will choose a node with a short distance.

[0123] 2. Pheromone update strategy. The formula is as follows:

[0124] τ ij (t+1)=(1-ρ)τij (t)+Δτ ij

[0125]

[0126] where τ ij (t) is the pheromone concentration from node i to node j at the current iteration, τ ij (t+1) is the pheromone concentration from node i to node j in the next iteration, is the pheromone released by the kth ant on the path from node i to node j, ρ is the pheromone attenuation coefficient, which represents the rate of pheromone volatilization on the path, L k is the length of the path taken by the kth ant, and Q is the total amount of pheromones released by the ant during a search process.

[0127] Improved ant colony algorithm improvements:

[0128] 1. Improvement of initial pheromone concentration. The formula is as follows:

[0129]

[0130] in, is the distance between the starting node and the target node, and The distances between the starting node and the current node, and between the current node and the candidate node, are shown in Figure 2. Nodes on the line connecting the starting and target nodes have higher pheromone concentrations, while nodes farther from the line have lower pheromone concentrations. This initial pheromone setting method enables the ant colony to search towards the target node at the beginning of the iteration, thereby improving search efficiency.

[0131] 2. Elite strategy: Add additional pheromones to the optimal path of the current iteration. The pheromone update formula is changed to the following:

[0132]

[0133] Among them, e is the elite coefficient, Is the elite ant on the path (L i ,L j ) increases the amount of pheromones released, Node L at the current iteration i To node L j The pheromone concentration, Node L for the next iteration i To node L j The pheromone concentration, For the kth ant at node L i To node L jThe pheromone released on the path, ρ is the pheromone attenuation coefficient, which represents the rate of pheromone volatilization on the path.

[0134] 3.2-opt method local path improvement.

[0135] Perform 2-opt optimization on the current optimal solution of each iteration to solve the problem of intersection of generated paths. If there are two local paths in the current optimal solution path of a certain iteration, they are: L i →L i +1 and L j →L j +1 if the exchange creates a new path L i →L j and the new path L i +1→L j +1, and meets: That is, if the length of the new path is shorter, it is replaced.

[0136] Step 4 realizes the collaborative path planning of the drone cluster and the unmanned boat under multiple constraints to ensure the efficiency and feasibility of the inspection mission.

[0137] Step 5: Select the partition with the shortest time among the feasible partitions

[0138] After completing the verification and calculation of all possible partitioning schemes, select K The minimum sub-area division scheme and the paths of UAVs and boats are the optimal collaborative path planning results in the absence of obstacles.

[0139] Step 6: Dealing with obstacles at the unmanned boat anchorage and planning the obstacle avoidance path in actual situations

[0140] 6.1 What to do if there are obstacles at the unmanned boat anchorage

[0141] If the unmanned boat is located at the planned berth point C in a sub-area under the optimal solution k If there is an obstacle at the stop point, this sub-area division scheme is directly abandoned, and the second best division scheme is found among the feasible sub-area division schemes when there is no obstacle. If there is still an obstacle at the stop point of the division scheme, the third best division scheme is continued to be found, and so on, until a sub-area division scheme that meets the requirements is found.

[0142] If the unmanned boat is at the planned berthing point C in a certain sub-area k If there is no obstacle, go to step 6.2.

[0143] 6.2 Unmanned Boat Obstacle Avoidance Path Planning

[0144] For the sub-area division scheme found, based on the overall path of the unmanned boat planned without obstacles, an improved ant colony algorithm with obstacle avoidance function is used to plan the overall obstacle avoidance path.

[0145] The obstacle repulsion term is introduced into the heuristic function:

[0146]

[0147] in, is node L j To the target node L g The Euclidean distance, λ is the weight coefficient, F r is the obstacle repulsion term, f is the number of obstacles within the field of view, d k is the distance from node j to the kth obstacle. The closer the obstacle is, the greater the repulsion.

[0148] In this step, obstacle handling at parking points uses "feasibility pruning" and "priority reuse" mechanisms to quickly identify viable solutions and avoid inefficient computations. The obstacle avoidance path planning component, based on the heuristic search properties of the ant colony algorithm, uses a physical model (repulsion term) to transform obstacle constraints into quantitative metrics that the algorithm can handle, balancing path feasibility and efficiency in complex environments.

[0149] Finally, the sub-area division scheme determined in step 6.1 and the final UAV and boat paths are selected as the optimal collaborative inspection path planning results in actual situations.

[0150] Experimental verification

[0151] To validate the proposed method for autonomous wind farm inspection path planning for UAV-boat collaboration based on sub-region partitioning and an improved ant colony algorithm, a simulation platform was constructed. Using parameters close to those of an actual medium-sized offshore wind farm, the feasibility and superiority of the proposed method were fully verified. The experiment evaluated the overall effectiveness of the UAV-boat collaborative path planning and the performance of the improved ant colony algorithm. The results demonstrated that the proposed method can significantly improve offshore wind farm inspection efficiency.

[0152] The simulation experiment design is divided into the following contents: (1) Parameter setting: The inspection area parameters are set based on the actual scale of the offshore wind farm. The parameters of the drone, such as inspection height, field of view, flight speed, and battery capacity, adopt the parameters of mainstream equipment; (2) Algorithm evaluation: The difference between the improved ant colony algorithm and the traditional ant colony algorithm in terms of convergence speed and path planning quality is compared; (3) Testing under ideal and actual conditions: Obstacles of different numbers and positions are randomly set in the simulation area to test the adaptability of the method. The simulation is built based on MATLAB, and the path planning results are evaluated by the total task completion time.

[0153] Overall process reference Figure 1 .

[0154] Combine Figure 2 , demonstrating the simulation results of the proposed method in an ideal obstacle-free environment. The 8km×8km square area in the figure represents the offshore wind farm to be inspected. The black dashed line in the XY plane represents the optimal navigation path for the unmanned boat between the centers of each sub-area. The black line group at a height of 200 meters above the first sub-area represents the flight paths of multiple drones within their respective sub-areas. As can be seen from the figure, based on the sub-area partitioning strategy, the entire wind farm is divided into 5×5=25 sub-areas. Task points within each sub-area are evenly distributed to multiple drones, enabling the drones to complete the inspection mission within the area while meeting endurance constraints. The maximum actual flight path length of the three drones within a single sub-area is 6526 meters. Furthermore, the unmanned boat's navigation path achieves the shortest path to all sub-area centers, with a total path length of 40700.08 meters and a total mission completion time of 32813.29 seconds. Simulation results demonstrate that compared to traditional inspection methods, the proposed method can fully cover the entire wind farm area, eliminating the need for drones to frequently return to the power station for recharging, significantly improving inspection efficiency.

[0155] Combine Figure 3 , shows the simulation results of the method of the present invention in an actual obstacle environment. In the figure, the gray blocks represent fixed obstacles (such as no-fly zones, offshore facilities, etc.), 80 obstacles are set, and the remaining elements are the same as Figure 2 The same can be observed. When obstacles are present, the unmanned boat's navigation path successfully plans the optimal path to avoid all obstacles by introducing an improved ant colony algorithm with an obstacle repulsion term. The path quality is high, balancing safety and smoothness. The maximum actual flight path length of the three drones in a single sub-area is 6526 meters, the total length of the unmanned boat path is 43081.74 meters, and the total time to complete the mission is 32510.67 seconds. This result verifies the adaptability and robustness of the method in complex offshore environments, and its ability to handle various obstacles that may exist in actual wind farms.

[0156] Combine Figure 4, demonstrating a comparison of the convergence performance of the improved ant colony algorithm proposed in this invention and the traditional ant colony algorithm for path planning. 30 random task points were generated in a 2600m x 1700m ocean area, and the two algorithms each planned a path that traversed all the task points. The horizontal axis represents the number of iterations, and the vertical axis represents the length of the currently found optimal path. The solid black line represents the convergence process of the traditional ant colony algorithm, while the dashed black line represents the convergence process of the improved ant colony algorithm proposed in this invention. The figure clearly shows that the improved ant colony algorithm quickly finds a better solution in the initial stage, converging significantly faster than the traditional algorithm. As the iterations proceed, the improved algorithm continuously optimizes the path, ultimately achieving a path length significantly shorter than that of the traditional algorithm. Specifically, the traditional ant colony algorithm found the optimal solution after 30 iterations, while the improved algorithm did so after 9 iterations, reducing the number of iterations by 70%. The shortest path length obtained by the traditional ant colony algorithm was 10499.6 meters, while the shortest path length obtained by the improved ant colony algorithm was 9915.83 meters, approximately 6% shorter than the traditional algorithm. This result fully verifies that the improvements in the initial pheromone setting, elite strategy and 2-opt local optimization of the proposed method significantly enhance the performance of the ant colony algorithm, providing a more efficient path planning solution for offshore wind farm inspections.

Claims

1. A method for autonomous inspection path planning for offshore wind farms using a UAV-boat collaboration. The method is based on the following known quantities: Parameters of the area to be inspected: length of the area to be inspected X max , Width of the area to be inspected Y max ; Rotor drone parameters: number of rotor drones m, rotor drone uniform flight speed v_a, rotor drone full charge E, rotor drone uniform charging speed β + , rotor UAV uniform discharge speed β - , Rotor UAV inspection flight altitude h, Optical sensor field of view angle Unmanned boat parameters: Unmanned boat speed v_s; Ant colony algorithm parameters: number of ants R, pheromone importance factor α, heuristic function importance factor β, pheromone attenuation coefficient ρ, total pheromone release Q, maximum number of iterations N_max; It is characterized by: The method comprises the following steps: Step 1: Determine the set of points to be inspected; Step 2: Divide the inspection area formed by the set of inspection points determined in step 1 into several inspection sub-areas; Step 3: For the Kth inspection sub-area partitioning scheme (K=1, 2, ..., N), assign drone tasks to the divided inspection sub-areas; Step 4: Use the improved ant colony algorithm ACO to plan the flight path of the UAV within the inspection sub-area and the navigation path of the UAV between the inspection sub-areas respectively; Step 5: Select the partitioning scheme with the shortest time among the feasible partitioning schemes; Step 6: Respond to obstacles at the unmanned boat berth and plan the obstacle avoidance path in actual situations.

2. The method for autonomous inspection path planning of an offshore wind farm using a UAV and a boat according to claim 1 is characterized in that: Step 1 specifically includes: Define the square sea surface area that can be effectively covered by the optical sensor carried by the rotor UAV at the inspection flight altitude as the "minimum inspection unit", and define the projection point of the center of the minimum inspection unit at the flight altitude h as the inspection task point of the rotor UAV. Assuming that the wind farm area to be inspected is a rectangular area, the rotor UAV flies at a fixed altitude h, and its optical sensor is vertically downward, with a field of view angle of The side length d of the minimum inspection unit is calculated by the following formula: Define the total number of minimum inspection units as M, Indicates the minimum number of inspection units in the length direction. Indicates the minimum number of inspection units in the width direction, then: The projection points of the center points of these inspection minimum units at the flight height h constitute the UAV inspection task point set of the wind farm. Let the center coordinates of the jth inspection minimum unit be p j =(x j ,y j ,0),j=1,2,…,M, thus obtaining the task point set P that the rotor UAV cluster needs to inspect j =(x j ,y j ,h),j=1,2,…,M。 3. The method for autonomous inspection path planning of an offshore wind farm using a UAV and a boat according to claim 2 is characterized in that: Step 2 is to divide each sub-area into the same size. First, consider the minimum number of inspection units in the length and width directions of the area to be inspected. and Factor, assuming There are a total of factors of except 1, namely S1, S2, ..., S a ; There are b factors of excluding 1, namely f1, f2, ..., f b By permuting and combining these factors, theoretically, N=a×b different sub-region size specifications can be generated, where the sub-region corresponding to the Kth partitioning scheme contains s w ×f r The minimum inspection unit is w=1,2,…,a; r=1,2,…,b.

4. The method for autonomous inspection path planning of an offshore wind farm using a UAV and a boat according to claim 3 is characterized in that: The specific method of step 3 is as follows: For the Kth partitioning scheme, K = 1, 2, ..., N, the UAV tasks are assigned to the divided sub-areas. The rotorcraft UAV has endurance constraints and must meet the following conditions: D≤D max Among them, E is the total power of the rotor UAV, v a is the uniform flight speed of the rotor UAV, β - is the uniform discharge speed of the rotor UAV, D max is the maximum distance that a rotary-wing UAV can fly with full power, and D is the maximum flight path length of a single rotary-wing UAV performing a mission in a sub-area; For a certain inspection sub-area, multiple drones conduct coordinated inspections. In order to make the time for each drone to perform tasks in the sub-area similar, thereby reducing the waiting time of the unmanned boat, the task allocation is based on the task point and the sub-area center, that is, the unmanned boat berthing point C. k The angle of the projection of the connecting line on the horizontal plane is set as the center point of the sub-area C k The coordinates of (x C ,y C ,0), the coordinates of a patrol task point A are (x A ,y A ,h); calculate the angle θ formed by the projection A′C of the line segment AC on the ground and the counterclockwise rotation of the positive direction of the X-axis; if 2(l-1)π / m≤θ<2lπ / m, l=1,2,…,m, then the inspection task point is assigned to the lth UAV.

5. The method for planning a path for autonomous inspection of an offshore wind farm using a drone and a boat according to claim 4 is characterized in that: Step 4 specifically includes the following sub-steps: 4.

1. UAV inspection path planning: For each UAV u, the improved ant colony algorithm is used to plan its flight path at the inspection point set it is assigned to. The path starts from the UAV berthing point C k It takes off vertically to a height of h, traverses all the inspection points assigned to it, and finally returns to the unmanned boat berthing point C. k ; make 1 means that UAV u in the Kth sub-area flies from task point i to task point j, otherwise it is 0, N ku is the number of task points that UAV u needs to inspect in sub-area K. The constraints include: Ensure that each inspection task point is only visited once by the corresponding drone; Make sure the drone is moving from the center of the sub-area C k take off and eventually return; 4.

2. Partition feasibility verification: Calculate the length D of each planned drone path. If the path lengths D of all drones in the sub-area satisfy D≤D max , then the current sub-area division and the task allocation of the UAV cluster are feasible, traverse and verify all partitions and remove infeasible partitions; 4.

3. Unmanned Boat Path Planning: If the current zoning scheme is feasible, all sub-area centers to be inspected, i.e., the berthing points C of the unmanned boats, will be k The improved ant colony algorithm is used to plan the shortest navigation path for the unmanned boat to start from the first anchorage point, visit all sub-area anchorage points once and finally return to the starting point. 4.4 Calculate total task completion time For the currently feasible sub-area division schemes and the corresponding UAV-boat collaborative path planning results, the total completion time t of the inspection task in the entire wind farm area under each division scheme is calculated. K (K=1,2,3,…,N), the formula is as follows: Δe=tβ - Where n represents the total number of sub-areas divided in the Kth partitioning case; t represents the time required for the rotor UAV cluster to take off and land on the UAV in a sub-area (the flight time in each sub-area is the same); t i,i+1 represents the travel time of the unmanned boat between two adjacent sub-areas; t n,1 represents the time it takes for the UAV to return from the center of the last sub-area to the center of the first sub-area; Δe represents the actual maximum power consumption of the UAV; e1 represents the minimum power required for the UAV to take off; b + Indicates the charging speed of the drone; β - Indicates the discharge speed of the drone, β + Indicates the charging speed of the drone; The improved ant colony algorithm described in step 4.1 and step 4.3 is as follows:

1. Initial pheromone concentration improvement, the formula is as follows: in, is the distance between the starting node and the target node, and are the distances between the starting node and the current node, and between the current node and the node to be selected; 2. Elite strategy: Additional pheromones are added to the optimal path of the current iteration, and the pheromone update formula is changed to the following: Among them, e is the elite coefficient, Is the elite ant on the path (L i ,L j ) increases the amount of pheromones released, Node L at the current iteration i To node L j The pheromone concentration, Node L for the next iteration i To node L j The pheromone concentration, For the kth ant at node L i To node L j The pheromone released on the path, ρ is the pheromone attenuation coefficient, which represents the rate of pheromone volatilization on the path; 3. Perform 2-opt optimization on the current optimal solution of each iteration to solve the problem of intersection of generated paths. If there are two local paths in the current optimal solution path of a certain iteration, they are: L i →L i +1 and L j →L j +1 if the exchange creates a new path L i →L j and the new path L i +1→L j +1, and meets: That is, if the length of the new path is shorter, it is replaced.

6. The method for planning a path for autonomous inspection of an offshore wind farm using a drone and a boat according to claim 5, characterized in that: Step 6 specifically includes the following sub-steps: 6.

1. What to do if there are obstacles at the unmanned boat anchorage: If the unmanned boat is located at the planned berth point C in a sub-area under the optimal solution k If there is an obstacle at the stop point, then this sub-area division scheme is directly abandoned, and the second best division scheme is found among the feasible sub-area division schemes when there is no obstacle. If there is still an obstacle at the stop point of this division scheme, then the third best division scheme is searched, and so on, until a sub-area division scheme that meets the requirements is found; If the unmanned boat is at the planned berthing point C in a certain sub-area k If there is no obstacle, go to step 6.2; 6.

2. Unmanned Boat Obstacle Avoidance Path Planning: For the sub-area division scheme found, based on the overall path of the unmanned boat planned without obstacles, an improved ant colony algorithm with obstacle avoidance function is used to plan the overall obstacle avoidance path, as follows: The obstacle repulsion term is introduced into the heuristic function: in, is node L j To the target node L g The Euclidean distance, λ is the weight coefficient, F r is the obstacle repulsion term, g is the number of obstacles within the field of view, d k is the distance from node j to the kth obstacle. The closer the obstacle is, the greater the repulsion; Finally, the sub-area division scheme determined in step 6.1 and the final UAV and boat paths are selected as the optimal collaborative inspection path planning results in actual situations.

Images