Multi-unmanned aerial vehicle inspection task re-planning method under aircraft nest fault

Abstract

The invention discloses a multi-unmanned aerial vehicle inspection task re-planning method under a nest fault. The method comprises the following steps: acquiring a fault nest number, coordinates of all residual task points, coordinates of a standby nest and residual inspection path data of each inspection unmanned aerial vehicle; setting a landing point of the inspection unmanned aerial vehicle corresponding to the fault aircraft nest as a landing preparation aircraft nest; establishing and solving a single unmanned aerial vehicle inspection task re-planning model under a fixed starting point and a fixed ending point so as to re-plan an inspection path of the inspection unmanned aerial vehicle corresponding to the fault aircraft nest; and establishing a multi-unmanned aerial vehicle task re-planning model, designing an improved adaptive large-scale neighborhood search algorithm, and solving the multi-unmanned aerial vehicle task re-planning model to obtain a task re-planning path with the shortest inspection time. The multi-unmanned aerial vehicle task re-planning method can efficiently solve the multi-unmanned aerial vehicle task re-planning problem when the nest fault abnormal state occurs at any time, so that the multi-unmanned aerial vehicle inspection system can continuously perform stable autonomous operation when the nest fault sudden abnormal state occurs, and the disaster tolerance capability of the unattended inspection system is remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of multi-UAV task planning, and in particular relates to a multi-UAV inspection task replanning method under a machine nest failure based on an adaptive large-scale neighborhood search algorithm. Background Art

[0002] Drones can overcome the geographical limitations of manual inspections. Equipped with various sensors, they are widely used in a wide range of inspection scenarios. The execution sequence of drone inspection tasks is typically planned synchronously, combining inspection task points with the location of the landing pad. However, equipment problems are inevitable during operation. If a sudden landing pad failure occurs during an inspection, the remaining energy of the drone after completing the original inspection may not be enough to support its return to the alternate landing pad. Real-time replanning is required, combining the remaining inspection task points with the alternate landing pad location to determine a feasible inspection path within energy constraints.

[0003] At present, there is little research on multi-UAV mission replanning methods in existing technologies. The task redistribution model based on the auction mechanism is prone to combinatorial explosion problems in multi-UAV collaboration scenarios, and is computationally complex and inefficient. Existing research lacks robustness in risky environments and lacks a special processing mechanism for machine nest failures. Summary of the Invention

[0004] In view of the above-mentioned deficiencies in the existing technology, the purpose of the present invention is to provide a method for re-planning multi-UAV inspection tasks under machine nest failure, so as to solve the technical problem that the existing inspection task planning scheme is difficult to achieve efficient multi-UAV task re-planning under system anomalies caused by machine nest failure. The method of the present invention improves the emergency response capability of the multi-UAV inspection system when encountering abnormal machine nest failure conditions.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] The present invention provides a method for replanning multi-UAV inspection tasks under a machine nest failure, comprising the following steps:

[0007] 1) Obtain the faulty landing nest number, coordinates of all remaining mission points, coordinates of the alternate landing nest, and remaining inspection path data of each inspection drone;

[0008] 2) The landing point of the inspection UAV corresponding to the faulty nest is set as the alternate nest; a single UAV inspection task replanning model with fixed start and end points is established and solved to replan the inspection path of the inspection UAV corresponding to the faulty nest;

[0009] 3) Establish a multi-UAV task replanning model, design and improve the adaptive large-scale neighborhood search algorithm, and use the improved adaptive large-scale neighborhood search algorithm to solve the multi-UAV task replanning model to obtain the task replanning path with the shortest inspection time.

[0010] Furthermore, the step 2) specifically includes:

[0011] 21) Set the inspection drone landing point corresponding to the faulty nest as the alternate landing nest;

[0012] Set up a backup landing nest in the center of the inspection area to be used for landing the corresponding inspection drone after the inspection mission is completed in case of a nest failure during the inspection process;

[0013] 22) Establish and solve a single UAV inspection task replanning model with fixed start and end points to replan the inspection path of the inspection UAV corresponding to the faulty machine nest;

[0014] A single UAV inspection task replanning model with fixed starting and ending points is established. The model includes: decision variables X ij and decision variable Y ij , decision variable X ij Indicates whether the inspection drone flies from point i to point j, the decision variable Y ij Delete the variable for the connection;

[0015] The decision variables are as follows:

[0016]

[0017] The re-planning phase of the single UAV inspection mission ignores the turning time and establishes an objective function with the goal of minimizing the direct flight time, as follows:

[0018]

[0019] The constraints are as follows:

[0020]

[0021] In the formula, T represents the inspection time of the inspection drone, T ij T represents the time required for the inspection UAV to fly from the inspection waypoint i to the inspection waypoint j. dj T represents the time required for the inspection UAV to fly from the current position d to the waypoint j to be inspected. ib represents the time required for the inspection UAV to fly from the inspection waypoint i to the coordinate position of the alternate landing nest b, Θ is the inspection task point set, and b is the alternate landing nest; u i 、u j They represent the order in which the inspection drone visits point i and point j, is a set of positive real numbers; Formula (4) indicates that for each waypoint to be inspected, the inspection drone enters exactly once; Formula (5) indicates that for each waypoint to be inspected, the inspection drone leaves exactly once; Formula (6) indicates that only one connection in the model is deleted, and its two endpoints are connected to the starting point and the alternate landing nest respectively; Formula (7) indicates that the deleted connection must not be composed of the same vertex; Formula (8) indicates that only existing connections can be deleted; Formulas (9) and (10) are MTZ (Miller-Tucker-Zemlin) sub-loop elimination constraints; Use CPLEX or gurobi solver to solve the single UAV inspection task replanning model with fixed starting and ending points, and obtain the inspection drone path S′ corresponding to the faulty landing nest with the shortest direct flight time. f .

[0022] Furthermore, the step 3) specifically includes:

[0023] 31) Establish a multi-UAV inspection task re-planning model as follows:

[0024]

[0025] Where, is an empty set; V represents the set of inspection drones; T k represents the inspection time of the kth inspection drone, including direct flight time and turning time; T max It represents the maximum inspection time of all inspection drones, that is, the time required to complete the remaining inspection tasks; S k It represents the remaining inspection path of the kth inspection drone, that is, the subsequent inspection paths that have not been completed by each inspection drone when the abnormal state of the machine nest failure occurs, among which the inspection path S of the inspection drone corresponding to the faulty machine nest is f Replaced by the inspection path S′ obtained in step 2) f , the inspection paths of other inspection drones are the remaining inspection paths of each inspection drone obtained in step 1); Represents the remaining inspection path S of the k-th inspection drone k The a-th point in Indicates that the inspection drone is Fly to point Flight time; Indicates that three consecutive points are Time inspection drone at the point U represents the current position coordinate set of the inspection UAV; Q represents the coordinate set of the remaining task points; B represents the coordinate set of the machine nest; Formula (13) indicates that all the remaining task points are visited by the inspection UAV, Formula (14) indicates that the inspection UAV cannot repeatedly visit the same remaining task point, S g、S l They represent the remaining inspection paths of the g-th and l-th inspection drones respectively;

[0026] 32) Design and improve the destruction operator of the adaptive large-scale neighborhood search algorithm as follows:

[0027] 321) Design and improve the nearest outlier destruction operator of the adaptive large-scale neighborhood search algorithm;

[0028] For each inspection drone in the set V, the average value of the coordinates of all remaining task points assigned to the inspection drone is calculated as the center position of the inspection drone task group, which is regarded as the dimensionality reduction representation of the coordinates of each waypoint in the task group. The task center of inspection drone k is denoted as C k ; The nearest outer class distance of the mth waypoint to be inspected belonging to the pth inspection drone task group for:

[0029]

[0030] Where, Respectively represent the waypoints to be inspected Q m 、Task Center C k The x-coordinate of Respectively represent the waypoints to be inspected Q m 、Task Center C k The y coordinate of Respectively represent the waypoints to be inspected Q m 、Task Center C k The z coordinate of

[0031] 322) Design and improve the perturbation nearest outlier destruction operator of the adaptive large-scale neighborhood search algorithm;

[0032] Use uniformly distributed noise ε to perturb the nearest outlier distance, ε∈[0.8,1.2], to expand the neighborhood search space; the perturbed nearest outlier distance of the mth waypoint to be inspected belonging to the pth inspection drone task group is for:

[0033]

[0034] 323) Design and improve the worst waypoint destruction operator of the adaptive large-scale neighborhood search algorithm;

[0035] For the mth waypoint to be inspected belonging to the pth inspection drone task group, set its traversal order in the current drone path to be nth; the removal benefit of the mth waypoint to be inspected belonging to the pth inspection drone task group is for:

[0036]

[0037] Where, They are respectively the n-1th and n+1th waypoints to be inspected in the traversal order of the pth inspection drone in the planned multi-drone inspection path, that is, the waypoints Q to be inspected m Directly connected waypoints;

[0038] 33) Design and improve the repair operator of the adaptive large-scale neighborhood search algorithm, as follows:

[0039] 331) Design and improve the greedy repair operator of the adaptive large-scale neighborhood search algorithm;

[0040] Calculate the change of the objective function when the waypoint to be repaired is inserted into each position of the solution, and insert the waypoint to be repaired into the position where the increase in inspection time is minimum; the repair cost between the vth waypoint to be repaired in the set of waypoints to be repaired H and the wth and w+1th waypoints to be inspected in the traversal order of the qth inspection drone is for:

[0041]

[0042] 332) Design and improve the regret value repair operator of the adaptive large-scale neighborhood search algorithm;

[0043] Calculate the regret value of all the waypoints to be inspected in the set H of waypoints to be repaired, and then select the waypoint with the largest regret value to be repaired until all the waypoints in set H are repaired. The regret value is the difference between the inspection time after the waypoint is inserted into the optimal position and the inspection time after it is inserted into the suboptimal position.

[0044] 333) Design and improve the perturbation repair operator of the adaptive large-scale neighborhood search algorithm;

[0045] Use uniformly distributed noise ε to perturb the repair cost or regret value of the damaged waypoints, and select the waypoint insertion position based on the perturbation result to obtain the perturbation greedy repair operator and the perturbation regret value repair operator to enhance the global optimization ability of the algorithm;

[0046] 34) Design and improve the acceptance criteria of the adaptive large-scale neighborhood search algorithm;

[0047] When searching for a solution using the destruction operator and the repair operator, determine whether the objective function is optimized during the iteration. If the objective function of the multi-UAV path replanning model is optimized during the iteration, the new solution is accepted and the next iteration is entered. If the objective function is not optimized during the iteration, the new solution is accepted and the next iteration is entered if one of the following conditions is met:

[0048] 341) The new solution is better than the current solution, but the sum of the inspection time of each drone is worse than the current solution;

[0049] 342) The new solution is not better than the current solution, but the increase in inspection time is less than the decrease in the sum of inspection times;

[0050] If none of the above conditions are met, the simulated annealing acceptance criterion is used; the probability prob of accepting an inferior solution under the simulated annealing acceptance criterion is calculated as follows:

[0051]

[0052] Where, T cur represents the objective function value of the current solution, i.e., the inspection time; T new represents the inspection time of the new solution; temp represents the temperature, which is cooled after each iteration using the formula temp = temp·c, where c is the cooling coefficient;

[0053] 35) Use the improved adaptive large-scale neighborhood search algorithm to solve the multi-UAV inspection task replanning model; the specific steps are as follows:

[0054] 351) Initialize the operator set and operator weights, where the destruction operator set includes the nearest outlier destruction operator, the perturbation nearest outlier destruction operator, the worst waypoint destruction operator, and the random destruction operator; the repair operator set includes the greedy repair operator, the perturbation greedy repair operator, the regret value repair operator, and the perturbation regret value repair operator; set the operator score, the maximum number of iterations, and the maximum number of no-improvement iterations;

[0055] 352) Randomly select an inspection path from the inspection paths other than the inspection path with the shortest time, and select a destruction operator to perform a destruction operation based on the operator weight;

[0056] 353) Selecting a repair operator based on the operator weight to generate a new solution;

[0057] 354) Determine whether to accept the new solution according to the acceptance criteria, and update the operator weight and simulated annealing temperature;

[0058] 355) Repeat steps 352) to 354) until the iteration end condition is reached, and the task replanning path with the shortest inspection time is obtained.

[0059] When a nest failure occurs during the UAV inspection process, a single UAV inspection task replanning model with fixed start and end points is first established for the UAV corresponding to the failed nest using equations (1)-(10), and the model is solved using CPLEX or gurobi solver to obtain the inspection path of the UAV corresponding to the failed nest. This inspection path and the remaining inspection paths of other inspection UAVs are then used as the initial solution. With the goal of minimizing the inspection time, an improved adaptive large-scale neighborhood search algorithm is used to replan the multi-UAV inspection tasks, and an effective multi-UAV inspection path under the abnormal nest failure is obtained.

[0060] Beneficial effects of the present invention:

[0061] Aiming at the problem of nest failure that may be encountered during multi-UAV inspection, the present invention sets the landing position of the UAV corresponding to the faulty nest as the alternate nest, and establishes a single UAV inspection task replanning model with fixed starting and ending points based on the current position of the UAV, the distribution of remaining task points and the position of the alternate nest. In addition, the invention designs a destruction operator, a repair operator and an acceptance criterion of an improved adaptive large-scale neighborhood search algorithm. The algorithm has strong global search capability, robustness and scalability, and can efficiently solve the problem of multi-UAV task replanning when an abnormal nest failure occurs at any time, so that the multi-UAV inspection system can continue to perform stable autonomous operations when encountering sudden abnormal nest failures, significantly improving the disaster recovery capability of the unmanned inspection system. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 Flowchart of the method of the present invention.

[0063] Figure 2 This is the initial state and the original inspection path in the embodiment of the present invention.

[0064] Figure 3 Schematic diagram of connection deletion variables established by the present invention.

[0065] Figure 4 This is the multi-UAV inspection path planned in the event of a machine nest failure according to an embodiment of the present invention. DETAILED DESCRIPTION

[0066] In order to facilitate understanding by those skilled in the art, the present invention will be further described below with reference to embodiments and drawings. The contents mentioned in the embodiments are not intended to limit the present invention.

[0067] Reference Figure 1 As shown, the present invention provides a method for replanning multi-UAV inspection tasks under a nest failure, and the steps are as follows:

[0068] 1) Obtain the faulty landing nest number, coordinates of all remaining mission points, coordinates of the alternate landing nest, and remaining inspection path data of each inspection drone;

[0069] The acquired data are marked as: fault nest number f, the current position u of the corresponding drone and the set of inspection task points Θ, the alternate nest number b and its coordinates, the current position coordinate set U of each drone, the coordinate set Q of the remaining task points, the nest coordinate set B, the remaining inspection path S of each drone, f∈B, u∈U, Θ∈Q; the data examples are shown in Table 1, the positions of each drone and the landing nest when the nest failure occurs, the distribution of the remaining task points and the original inspection path are as follows Figure 2 As shown;

[0070] Table 1

[0071]

[0072]

[0073] 2) Set the landing point of the inspection drone corresponding to the faulty nest as the alternate nest; establish and solve a single drone inspection task replanning model with fixed start and end points to replan the inspection path of the inspection drone corresponding to the faulty nest; specifically,

[0074] 21) Set the inspection drone landing point corresponding to the faulty nest as the alternate landing nest;

[0075] Set up a backup landing nest in the center of the inspection area to be used for landing the corresponding inspection drone after the inspection mission is completed in case of a nest failure during the inspection process;

[0076] 22) Establish and solve a single UAV inspection task replanning model with fixed start and end points to replan the inspection path of the inspection UAV corresponding to the faulty machine nest;

[0077] A single UAV inspection task replanning model with fixed starting and ending points is established. The model includes: decision variables X ij and decision variable Y ij , decision variable X ij Indicates whether the inspection drone flies from point i to point j, the decision variable Y ij Delete variable for connection; Y ij The inspection drone path when the value is 1 is as follows Figure 3 As shown;

[0078] The decision variables are as follows:

[0079]

[0080] The re-planning phase of the single UAV inspection mission ignores the turning time and establishes an objective function with the goal of minimizing the direct flight time, as follows:

[0081]

[0082] The constraints are as follows:

[0083]

[0084] In the formula, T represents the inspection time of the inspection drone, T ij T represents the time required for the inspection UAV to fly from the inspection waypoint i to the inspection waypoint j. dj T represents the time required for the inspection UAV to fly from the current position d to the waypoint j to be inspected. ib represents the time required for the inspection UAV to fly from the inspection waypoint i to the coordinate position of the alternate landing nest b, Θ is the inspection task point set, and b is the alternate landing nest; u i 、u j They represent the order in which the inspection drone visits point i and point j, is a set of positive real numbers; Formula (4) indicates that for each waypoint to be inspected, the inspection drone enters exactly once; Formula (5) indicates that for each waypoint to be inspected, the inspection drone leaves exactly once; Formula (6) indicates that only one connection in the model is deleted, and its two endpoints are connected to the starting point and the alternate landing nest respectively; Formula (7) indicates that the deleted connection must not be composed of the same vertex; Formula (8) indicates that only existing connections can be deleted; Formulas (9) and (10) are MTZ (Miller-Tucker-Zemlin) sub-loop elimination constraints; Use CPLEX or gurobi solver to solve the single UAV inspection task replanning model with fixed starting and ending points, and obtain the inspection drone path S′ corresponding to the faulty landing nest with the shortest direct flight time. f .

[0085] Use gurobi or CPLEX to solve the model and obtain the inspection path of the drone corresponding to the faulty machine nest: (1332,859,1185),(1127,870,1254),(924,741,1193),...,(1107,1066,1240),(1573,1964,888).

[0086] 3) Establish a multi-UAV mission replanning model, design and improve the adaptive large-scale neighborhood search algorithm, and use the improved adaptive large-scale neighborhood search algorithm to solve the multi-UAV mission replanning model to obtain the mission replanning path with the shortest inspection time; specifically,

[0087] 31) Establish a multi-UAV inspection task re-planning model as follows:

[0088]

[0089] Where V represents the set of inspection drones; T krepresents the inspection time of the kth inspection drone, including direct flight time and turning time; T max It represents the maximum inspection time of all inspection drones, that is, the time required to complete the remaining inspection tasks; S k It represents the remaining inspection path of the kth inspection drone, that is, the subsequent inspection paths that have not been completed by each inspection drone when the abnormal state of the machine nest failure occurs, among which the inspection path S of the inspection drone corresponding to the faulty machine nest is f Replaced by the inspection path S′ obtained in step 2) f , the inspection paths of other inspection drones are the remaining inspection paths of each inspection drone obtained in step 1); Represents the remaining inspection path S of the k-th inspection drone k The a-th point in Indicates that the inspection drone is Fly to point Flight time; Indicates that three consecutive points are Time inspection drone at the point U represents the current position coordinate set of the inspection UAV; Q represents the coordinate set of the remaining task points; B represents the coordinate set of the machine nest; Formula (13) indicates that all the remaining task points are visited by the inspection UAV, Formula (14) indicates that the inspection UAV cannot repeatedly visit the same remaining task point, S g 、S l They represent the remaining inspection paths of the g-th and l-th inspection drones respectively;

[0090] 32) Design and improve the destruction operator of the adaptive large-scale neighborhood search algorithm as follows:

[0091] 321) Design and improve the nearest outlier destruction operator of the adaptive large-scale neighborhood search algorithm;

[0092] For each inspection drone in the set V, the average value of the coordinates of all remaining task points assigned to the inspection drone is calculated as the center position of the inspection drone task group, which is regarded as the dimensionality reduction representation of the coordinates of each waypoint in the task group. The task center of inspection drone k is denoted as C k ; The nearest outer class distance of the mth waypoint to be inspected belonging to the pth inspection drone task group for:

[0093]

[0094] Where, Respectively represent the waypoints to be inspected Q m 、Task Center C k The x-coordinate of Respectively represent the waypoints to be inspected Q m 、Task Center C k The y coordinate of Respectively represent the waypoints to be inspected Q m 、Task Center C k The z coordinate of

[0095] 322) Design and improve the perturbation nearest outlier destruction operator of the adaptive large-scale neighborhood search algorithm;

[0096] Use uniformly distributed noise ε to perturb the nearest outlier distance, ε∈[0.8,1.2], to expand the neighborhood search space; the perturbed nearest outlier distance of the mth waypoint to be inspected belonging to the pth inspection drone task group is for:

[0097]

[0098] 323) Design and improve the worst waypoint destruction operator of the adaptive large-scale neighborhood search algorithm;

[0099] For the mth waypoint to be inspected belonging to the pth inspection drone task group, set its traversal order in the current drone path to be nth; the removal benefit of the mth waypoint to be inspected belonging to the pth inspection drone task group is for:

[0100]

[0101] Where, They are respectively the n-1th and n+1th waypoints to be inspected in the traversal order of the pth inspection drone in the planned multi-drone inspection path, that is, the waypoints Q to be inspected m Directly connected waypoints;

[0102] 33) Design and improve the repair operator of the adaptive large-scale neighborhood search algorithm, as follows:

[0103] 331) Design and improve the greedy repair operator of the adaptive large-scale neighborhood search algorithm;

[0104] Calculate the change of the objective function when the waypoint to be repaired is inserted into each position of the solution, and insert the waypoint to be repaired into the position where the increase in inspection time is minimum; the repair cost between the vth waypoint to be repaired in the set of waypoints to be repaired H and the wth and w+1th waypoints to be inspected in the traversal order of the qth inspection drone is for:

[0105]

[0106] 332) Design and improve the regret value repair operator of the adaptive large-scale neighborhood search algorithm;

[0107] Calculate the regret value of all the waypoints to be inspected in the set H of waypoints to be repaired, and then select the waypoint with the largest regret value to be repaired until all the waypoints in set H are repaired. The regret value is the difference between the inspection time after the waypoint is inserted into the optimal position and the inspection time after it is inserted into the suboptimal position.

[0108] 333) Design and improve the perturbation repair operator of the adaptive large-scale neighborhood search algorithm;

[0109] Use uniformly distributed noise ε to perturb the repair cost or regret value of the damaged waypoints, and select the waypoint insertion position based on the perturbation result to obtain the perturbation greedy repair operator and the perturbation regret value repair operator to enhance the global optimization ability of the algorithm;

[0110] 34) Design and improve the acceptance criteria of the adaptive large-scale neighborhood search algorithm;

[0111] When searching for a solution using the destruction and repair operators, determine whether the objective function is optimized during the iteration. If the objective function of the multi-UAV path replanning model is optimized during the iteration, the new solution is accepted and the next iteration is entered. If the objective function is not optimized during the iteration, the new solution is accepted and the next iteration is entered if one of the following conditions is met:

[0112] 341) The new solution is better than the current solution, but the sum of the inspection time of each drone is worse than the current solution;

[0113] 342) The new solution is not better than the current solution, but the increase in inspection time is less than the decrease in the sum of inspection times;

[0114] If none of the above conditions are met, the simulated annealing acceptance criterion is used; the probability prob of accepting an inferior solution under the simulated annealing acceptance criterion is calculated as follows:

[0115]

[0116] Where, T cur represents the objective function value of the current solution, i.e., the inspection time; T new represents the inspection time of the new solution; temp represents the temperature, which is cooled after each iteration using the formula temp = temp·c, where c is the cooling coefficient;

[0117] 35) Use the improved adaptive large-scale neighborhood search algorithm to solve the multi-UAV inspection task replanning model; the specific steps are as follows:

[0118] 351) Initialize the operator set and operator weights, where the destruction operator set includes the nearest outlier destruction operator, the perturbation nearest outlier destruction operator, the worst waypoint destruction operator, and the random destruction operator; the repair operator set includes the greedy repair operator, the perturbation greedy repair operator, the regret value repair operator, and the perturbation regret value repair operator; set the operator score, the maximum number of iterations, and the maximum number of no-improvement iterations;

[0119] 352) Randomly select an inspection path from the inspection paths other than the inspection path with the shortest time, and select a destruction operator to perform a destruction operation based on the operator weight;

[0120] 353) Selecting a repair operator based on the operator weight to generate a new solution;

[0121] 354) Determine whether to accept the new solution according to the acceptance criteria, and update the operator weight and simulated annealing temperature;

[0122] 355) Repeat steps 352) to 354) until the iteration end condition is reached, and the task replanning path with the shortest inspection time is obtained.

[0123] In this example, the objective function value is 843.09s, and the task execution times of the four drones are 809.17s, 839.40s, 843.09s, and 806.27s respectively. The inspection paths of each drone are:

[0124] S 1 :((1332,859,1185),(1139,693,1256),(1127,870,1254),...,(924,741,1193),(777,974,987));

[0125] S 2 :((2673,707,1119),(2646,889,1130),(2584,1063,1131),...,(2107,107,1177),(2367,978,866));

[0126] S 3 :((899,3019,1130),(1090,3022,1091),(1075,2880,1105),...,(652,3366,1018),(851,2909,927));

[0127] S 4:((2880,3065,1134),(2634,2649,1006),(2180,3107,1056),...,(1600,2872,983),(2476,2998,874));

[0128] The re-planned inspection path is as follows Figure 4 shown.

[0129] The present invention has many specific application paths. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements can be made without departing from the principles of the present invention. These improvements should also be considered as the scope of protection of the present invention.

Claims

1. A method for replanning multi-UAV inspection tasks under a machine nest failure, characterized in that: Here are the steps: 1) Obtain the faulty landing nest number, coordinates of all remaining mission points, coordinates of the alternate landing nest, and remaining inspection path data of each inspection drone; 2) The landing point of the inspection UAV corresponding to the faulty nest is set as the alternate nest; a single UAV inspection task replanning model with fixed start and end points is established and solved to replan the inspection path of the inspection UAV corresponding to the faulty nest; 3) Establish a multi-UAV task replanning model, design and improve the adaptive large-scale neighborhood search algorithm, and use the improved adaptive large-scale neighborhood search algorithm to solve the multi-UAV task replanning model to obtain the task replanning path with the shortest inspection time.

2. The method for replanning multi-UAV inspection tasks under a machine nest failure according to claim 1 is characterized in that: The step 2) specifically includes: 21) Set the inspection drone landing point corresponding to the faulty nest as the alternate nest; Set up a backup landing nest in the center of the inspection area to be used for landing the corresponding inspection drone after the inspection mission is completed in case of a nest failure during the inspection process; 22) Establish and solve a single UAV inspection task replanning model with fixed start and end points to replan the inspection path of the inspection UAV corresponding to the faulty machine nest; A single UAV inspection task replanning model with fixed starting and ending points is established. The model includes: decision variables X ij and decision variable Y ij , decision variable X ij Indicates whether the inspection drone flies from point i to point j, the decision variable Y ij Delete the variable for the connection; The decision variables are as follows: The re-planning phase of the single UAV inspection mission ignores the turning time and establishes an objective function with the goal of minimizing the direct flight time, as follows: The constraints are as follows: In the formula, T represents the inspection time of the inspection drone, T ij T represents the time required for the inspection UAV to fly from the inspection waypoint i to the inspection waypoint j. dj T represents the time required for the inspection UAV to fly from the current position d to the waypoint j to be inspected. ib represents the time required for the inspection UAV to fly from the inspection waypoint i to the coordinate position of the alternate landing nest b, Θ is the inspection task point set, and b is the alternate landing nest; u i 、u j They represent the order in which the inspection drone visits point i and point j, is a set of positive real numbers; use CPLEX or gurobi solver to solve the single UAV inspection task replanning model with fixed starting and ending points, and obtain the inspection UAV path S′ corresponding to the faulty machine nest with the shortest direct flight time f .

3. The method for replanning multi-UAV inspection tasks under a machine nest failure according to claim 2 is characterized in that: The step 3) specifically includes: 31) Establish a multi-UAV inspection task re-planning model as follows: Where, is an empty set; V represents the set of inspection drones; T k represents the inspection time of the kth inspection drone, including direct flight time and turning time; T max It represents the maximum inspection time of all inspection drones, that is, the time required to complete the remaining inspection tasks; S k It represents the remaining inspection path of the kth inspection drone, that is, the subsequent inspection paths that have not been completed by each inspection drone when the abnormal state of the machine nest failure occurs, among which the inspection path S of the inspection drone corresponding to the faulty machine nest is f Replaced by the inspection path S′ obtained in step 2) f , the inspection paths of other inspection drones are the remaining inspection paths of each inspection drone obtained in step 1); Represents the remaining inspection path S of the k-th inspection drone k The a-th point in Indicates that the inspection drone is Fly to point Flight time; Indicates that three consecutive points are Time inspection drone at the point Turn time; U represents the current position coordinate set of the inspection UAV; Q represents the coordinate set of the remaining task points; B represents the coordinate set of the machine nest; S g 、S l They represent the remaining inspection paths of the g-th and l-th inspection drones respectively; 32) Design and improve the destruction operator of the adaptive large-scale neighborhood search algorithm as follows: 321) Design and improve the nearest outlier destruction operator of the adaptive large-scale neighborhood search algorithm; For each inspection drone in the set V, the average coordinates of all remaining task points assigned to the inspection drone are calculated as the center position of the inspection drone task group, which is regarded as the dimensionality reduction representation of the coordinates of each waypoint in the task group. The task center of inspection drone k is denoted as C k ; The nearest outer class distance of the mth waypoint to be inspected belonging to the pth inspection drone task group for: Where, Respectively represent the waypoints to be inspected Q m 、Task Center C k The x-coordinate of Respectively represent the waypoints to be inspected Q m 、Task Center C k The y coordinate of Respectively represent the waypoints to be inspected Q m 、Task Center C k The z coordinate of 322) Design and improve the perturbation nearest outlier destruction operator of the adaptive large-scale neighborhood search algorithm; Use uniformly distributed noise ε to perturb the nearest outlier distance, ε∈[0.8,1.2], to expand the neighborhood search space; the perturbed nearest outlier distance of the mth waypoint to be inspected belonging to the pth inspection drone task group is for: 323) Design and improve the worst waypoint destruction operator of the adaptive large-scale neighborhood search algorithm; For the mth waypoint to be inspected belonging to the pth inspection drone task group, set its traversal order in the current drone path to be nth; the removal benefit of the mth waypoint to be inspected belonging to the pth inspection drone task group is for: Where, They are respectively the n-1th and n+1th waypoints to be inspected in the traversal order of the pth inspection drone in the planned multi-drone inspection path, that is, the waypoints Q to be inspected m Directly connected waypoints; 33) Design and improve the repair operator of the adaptive large-scale neighborhood search algorithm, as follows: 331) Design and improve the greedy repair operator of the adaptive large-scale neighborhood search algorithm; Calculate the change of the objective function when the waypoint to be repaired is inserted into each position of the solution, and insert the waypoint to be repaired into the position where the increase in inspection time is minimum; the repair cost between the vth waypoint to be repaired in the set of waypoints to be repaired H and the wth and w+1th waypoints to be inspected in the traversal order of the qth inspection drone is for: 332) Design and improve the regret repair operator of the adaptive large-scale neighborhood search algorithm; Calculate the regret value of all the waypoints to be inspected in the set H of waypoints to be repaired, and then select the waypoint with the largest regret value to be repaired until all the waypoints in set H are repaired. The regret value is the difference between the inspection time after the waypoint is inserted into the optimal position and the inspection time after it is inserted into the suboptimal position. 333) Design and improve the perturbation repair operator of the adaptive large-scale neighborhood search algorithm; Use uniformly distributed noise ε to perturb the repair cost or regret value of the damaged waypoints, and select the waypoint insertion position based on the perturbation result to obtain the perturbation greedy repair operator and the perturbation regret value repair operator; 34) Design and improve the acceptance criteria of the adaptive large-scale neighborhood search algorithm; When searching for a solution using the destruction operator and the repair operator, determine whether the objective function is optimized during the iteration. If the objective function of the multi-UAV path replanning model is optimized during the iteration, the new solution is accepted and the next iteration is entered. If the objective function is not optimized during the iteration, the new solution is accepted and the next iteration is entered if one of the following conditions is met: 341) The new solution is better than the current solution, but the sum of the inspection time of each drone is worse than the current solution; 342) The new solution is not better than the current solution, but the increase in inspection time is less than the decrease in the sum of inspection times; If none of the above conditions are met, the simulated annealing acceptance criterion is used; the probability prob of accepting an inferior solution under the simulated annealing acceptance criterion is calculated as follows: Where, T cur represents the objective function value of the current solution, i.e., the inspection time; T new represents the inspection time of the new solution; temp represents the temperature, which is cooled after each iteration using the formula temp = temp·c, where c is the cooling coefficient; 35) Use the improved adaptive large-scale neighborhood search algorithm to solve the multi-UAV inspection task replanning model; the specific steps are as follows: 351) Initialize the operator set and operator weights, where the destruction operator set includes the nearest outlier destruction operator, the perturbation nearest outlier destruction operator, the worst waypoint destruction operator, and the random destruction operator; the repair operator set includes the greedy repair operator, the perturbation greedy repair operator, the regret value repair operator, and the perturbation regret value repair operator; set the operator score, the maximum number of iterations, and the maximum number of no-improvement iterations; 352) Randomly select an inspection path from the inspection paths other than the inspection path with the shortest time, and select a destruction operator to perform a destruction operation based on the operator weight; 353) Selecting a repair operator based on the operator weight to generate a new solution; 354) Determine whether to accept the new solution according to the acceptance criteria, and update the operator weight and simulated annealing temperature; 355) Repeat steps 352) to 354) until the iteration end condition is reached, and the task replanning path with the shortest inspection time is obtained.

Images