Truck-unmanned aerial vehicle cooperative path optimization method and program product for heterogeneous inspection tasks

Abstract

The application discloses a truck-unmanned aerial vehicle cooperative path optimization method and program product for heterogeneous inspection tasks, and comprises the following steps: (1) setting a target function of a truck-unmanned aerial vehicle cooperative path optimization model as minimizing total cost, (2) adding route constraints, heterogeneity constraints of different types of inspection tasks on equipment, unmanned aerial vehicle energy consumption constraints and time constraints to the truck-unmanned aerial vehicle cooperative path optimization model; (3) solving the truck-unmanned aerial vehicle cooperative path optimization model to obtain optimized decision variables, so as to obtain an optimal route and an optimal unmanned aerial vehicle-truck configuration scheme for heterogeneous inspection tasks. The application considers the heterogeneity of the inspection tasks, and the planned path scheme has higher inspection efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent transportation and inspection, and particularly relates to a truck-unmanned aerial vehicle (UAV) collaborative path optimization method and program product for heterogeneous inspection tasks. BACKGROUND

[0002] With the advancement of UAV technology, truck-UAV collaborative systems have been applied in urban traffic patrol, environmental data collection, forest fire patrol and other fields to perform inspection tasks. In such systems, trucks can perform some ground tasks that require human participation, while serving as mobile platforms responsible for carrying, launching and recovering UAVs and providing battery replacement. UAVs can use their flight capabilities to reach target areas directly, collect high-resolution data, and perform inspection tasks in dangerous environments. By combining the advantages of trucks and UAVs, this system can efficiently inspect a wider geographical area, reducing labor costs while improving safety and efficiency.

[0003] In practical applications, inspection tasks are heterogeneous due to differences in inspection content, environment and other characteristics, and need to be performed by UAVs or trucks equipped with corresponding functional devices. For example, a task to detect fire risks may require a UAV equipped with an infrared sensor, while a facility inspection task may require a high-resolution camera or laser radar. Therefore, when planning, the truck-UAV collaborative inspection system needs to solve the problem of task adaptability and the allocation of multiple UAVs to trucks, jointly optimizing the truck fleet and task route to reduce overall costs and improve overall efficiency.

[0004] Currently, although there are a wide range of heterogeneous tasks in reality, most existing UAV inspection research focuses on uniform task settings and does not fully consider the diverse needs of different types of tasks, making it difficult to meet the needs of task heterogeneity. At the same time, existing research usually plans routes and allocates tasks with UAVs and trucks as fixed combinations, resulting in low inspection efficiency of the planned path scheme. SUMMARY

[0005] To solve the problems in the prior art, the present application aims to provide a truck-UAV collaborative path optimization method and program product for heterogeneous inspection tasks with higher inspection efficiency.

[0006] To achieve the above-mentioned application purpose, the present application provides the following technical solutions:

[0007] A truck-UAV collaborative path optimization method for heterogeneous inspection tasks, comprising the following steps:

[0008] (1) Set the objective function of the truck-UAV collaborative path optimization model as:

[0009] ,

[0010] Where, represents the total cost, They represent the truck startup cost and the unit time driving cost respectively. 、 They represent the startup cost and the unit time driving cost of the UAV respectively. 、 are all truck route variables, respectively in truck From warehouse to task node Driving, from the task node arrive The value is 1 when driving, otherwise it is 0. Configure variables for drone-truck, in drone The value is 1 if it is assigned to truck k, otherwise it is 0. 、 Represents truck and drone slave nodes respectively arrive driving time, is the drone route variable. Itinerary For slave task nodes i arrive The value is 1 when driving, otherwise it is 0. is the set of all task nodes except warehouse and terminal. Represents the set of all task nodes except the end point, Represents the set of all task nodes except the starting point, represents a collection of drones, represents the set of trucks, represents the set of drone itineraries, 、 、 、 is the decision variable;

[0011] (2) Add routing constraints, heterogeneity constraints on equipment for different types of inspection tasks, and UAV energy consumption and time constraints to the truck-UAV collaborative path optimization model;

[0012] (3) Solve the truck-UAV collaborative path optimization model to obtain the optimized decision variables, thereby obtaining the optimal route and optimal UAV-truck configuration solution for heterogeneous inspection tasks.

[0013] A computer program product comprises a computer program / instruction, wherein the computer program / instruction implements the above method when executed by a processor.

[0014] Compared with the prior art, the present invention has the following beneficial effects:

[0015] 1. For the problem of truck and UAV cooperation to perform heterogeneous inspection tasks, a mixed integer programming model is successfully constructed to accurately reflect the complexity of the task, which comprehensively considers the heterogeneity requirements of different types of inspection tasks on equipment, the battery endurance of UAVs, the formation of truck and UAV teams, and various constraint conditions such as task time, etc. The association between UAVs and trucks is not fixed as a matching but as an optimized parameter, which can effectively optimize resource allocation and task route in complex actual environment, complete all tasks with fewer number of UAVs and trucks, improve the flexibility and efficiency of inspection task execution, and significantly reduce the total cost.

[0016] 2. The application also proposes a self-adaptive large neighborhood search algorithm to solve the model parameters, which can flexibly cope with different size and type of task demand scenarios, quickly give high-quality schemes, and ensure the efficiency of solving. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The flowchart of the truck-UAV cooperative path optimization method for heterogeneous inspection tasks provided by the application is shown.

[0018] Figure 2 The architecture suitable for the method of the application is shown.

[0019] Figure 3 The flowchart of the truck-UAV cooperative path optimization model solving method of the application is shown.

[0020] Figure 4 The numerical experiment example result visualization diagram of the application is shown. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application.

[0022] Embodiment one

[0023] The embodiment provides a truck-UAV cooperative path optimization method for heterogeneous inspection tasks to solve the problem of UAV-truck cooperative inspection, as shown in the figure, the UAV-truck cooperative inspection problem is defined as follows: Figure 2

[0024] a) Each truck carries at least one UAV, and the truck has a maximum UAV carrying capacity limit;

[0025] b) There are many different types of UAVs that can perform different types of tasks;

[0026] ​c) Both truck and drone can patrol the task node, but need to meet the requirements of the node to the patrol device;

[0027] d) The truck not only undertakes the execution of the ground task, but also transports the drone and serves as a mobile take-off and landing platform for the drone, providing launching, recovery and battery supply for the drone;

[0028] e) Each drone can perform multiple tasks after one departure if the power allows;

[0029] f) If the drone arrives at the meeting point earlier than the truck, it can land at the point and wait for the truck to complete the recovery operation;

[0030] g) All vehicle teams start from the warehouse and return to the warehouse within the specified time window.

[0031] The assumptions of the problem addressed by the present application are as follows:

[0032] a) The truck and the drone are not allowed to visit the node repeatedly;

[0033] b) Each node must and can be visited only once;

[0034] c) After the drone returns to the truck, the battery will be replaced with a full battery.

[0035] For the above inspection problem and assumptions, as shown in Figure 1 , the method of the present embodiment comprises the following steps:

[0036] ,

[0037] In the formula, represents the total cost, respectively represent the truck start-up cost and the unit time driving cost, , respectively represent the drone start-up cost and the unit time driving cost, , are all truck driving route variables, respectively taking the value 1 when the truck drives from the warehouse to the task node and from the task node to , and 0 otherwise, is a drone-truck configuration variable, taking the value 1 when the drone is configured on the truck k, and 0 otherwise, , respectively represent the driving time of the truck and the drone from the node to , is a drone route variable, taking the value 1 when the drone is on the route For slave task nodes i arrive The value is 1 when driving, otherwise it is 0. is the set of all task nodes except warehouse and terminal. Represents the set of all task nodes except the end point, Represents the set of all task nodes except the starting point, represents a collection of drones, represents the set of trucks, represents the set of drone itineraries, 、 、 、 is the decision variable.

[0038] (2) Add routing constraints, heterogeneity constraints on equipment for different types of inspection tasks, and UAV energy consumption constraints and time constraints to the truck-UAV collaborative path optimization model.

[0039] Routing constraints include:

[0040] ,

[0041] ,

[0042] ,

[0043] ,

[0044] ,

[0045] ,

[0046] ,

[0047] ,

[0048] ,

[0049] ,

[0050] ,

[0051] ,

[0052] ,

[0053] ,

[0054] ,

[0055] ,

[0056] wherein, is a truck route variable, is a truck-task configuration variable, which is 1 when task node is visited by the truck and 0 otherwise, , denote the position of task node , j in the truck route, denotes the number of task nodes in, denotes the set of all task nodes, denotes a drone take-off point variable, which is 1 when the drone takes off from in trip f and 0 otherwise, denotes a drone take-off point variable, which is 1 when the drone takes off from in trip f+1 and 0 otherwise, denotes a drone take-up point variable, which is 1 when the drone is taken up at in trip f and 0 otherwise, denotes a drone take-up point variable, which is 1 when the drone is taken up at in trip f and 0 otherwise. are drone trip variables, which are 1 when task node , j is visited in trip f of the drone and 0 otherwise, denotes a drone route variable, denotes the maximum number of drones the truck can carry. is a truck-task configuration variable, which is 1 when task node is visited by the truck and 0 otherwise.

[0057] The first to fifth constraints above guarantee the feasibility of the truck path by balancing the flow, eliminating sub-loops, and ensuring that each node is visited exactly once. Similarly, the sixth to thirteenth constraints guarantee the feasibility of the drone trips. The fourteenth constraint ensures that each task is performed only once; the fifteenth constraint ensures that each edge traversed by the drone is associated with at least one assigned task; and the sixteenth constraint forces the drone trip numbers to be consecutive positive integers.

[0058] The heterogeneous constraints imposed by different types of inspection tasks on equipment include:

[0059] ,

[0060] ,

[0061] Where, Represents the node-truck inspection variable, when the task node Can be transported by truck The value is 1 when the inspection is in progress, otherwise it is 0. Indicates node-UAV inspection variable, when the task node Can be operated by drone d The value is 1 during inspection and 0 otherwise. This constraint ensures that each task node is assigned to a truck or drone with corresponding execution capabilities.

[0062] The energy consumption constraints of the UAV include:

[0063] ,

[0064] ,

[0065] ,

[0066] ,

[0067] ,

[0068] Where, 、 Respectively represent drones At the task node , the remaining power at time j, Indicates the drone battery capacity, Represents a task node Inspection time.

[0069] Of the constraints above, the first and second constraints ensure that each drone trip begins with a fully charged battery. The third and fourth constraints track the drone's battery level during the trip. The fifth constraint ensures that the drone has enough battery to complete the trip.

[0070] The time constraints include:

[0071] ,

[0072] ,

[0073] ,

[0074] ,

[0075] ,

[0076] ,

[0077] ,

[0078] ,

[0079] ,

[0080] ,

[0081] ,

[0082] ,

[0083] ,

[0084] ,

[0085] ,

[0086] , 、 , , , , , , , , , , , , , , , , , , , , , ,

[0087] Among the above constraints, the first and second constraints specify the departure time of all vehicle fleets. The third to eleventh constraints are used to track the arrival and departure time of trucks and drones at each visited node. The twelfth and thirteenth constraints determine the start time of each new drone trip. The fourteenth constraint ensures that the drone has been recovered before taking off again. The fifteenth constraint forces all trucks to return to the warehouse within the specified time window.

[0088] (3) Solving the truck-drones collaborative path optimization model to obtain the optimized decision variables, thereby obtaining the optimal route and optimal drone-truck configuration scheme for heterogeneous inspection tasks.

[0089] As shown in Figure 3 , this step uses an adaptive large neighborhood algorithm to solve the model, which specifically includes:

[0090] (3-1) Based on the preset initial truck-drones configuration scheme, use the nearest neighbor insertion method to construct the initial task execution route of the truck and the drone. After the route is constructed, remove all idle drones and return them to the warehouse.

[0091] First stage: according to the nearest neighbor insertion strategy, considering the time window constraint, construct the route of the truck executing the task;

[0092] Second stage: for each uninserted drone task point, find the nearest truck node. If the truck on this node has available drones, then this truck node will be used as the launch node, and the drone node will be immediately inserted after this node. The next truck node that satisfies the battery capacity constraint will be selected as the recovery node. If the nearest truck point cannot be inserted, consider the second nearest truck point, and so on.

[0093] When all points have been visited, the drone operation may affect the waiting time of the truck at each recovery node, thereby affecting the time of the truck returning to the warehouse. Therefore, some initial solutions may violate the time window constraint of the warehouse. At this time, it is allowed to use infeasible initial solutions, and the time window violation part is added as a penalty term to the objective function to be punished.

[0094] Set the number of iterations Iteration=0.

[0095] (3-2) Use the roulette strategy to select and apply the destruction operator and repair operator to generate a new route, and introduce or recover drones.

[0096] where the destroy operator and the repair operator perturb the current solution 9 (current route) to generate a new solution (new route). In the present invention, 7 destroy operators and 9 repair operators are proposed. The destroy operator is used to remove a portion of nodes from the current route, thus creating an incomplete route. The purpose of this is to be able to explore new neighborhoods in the subsequent repair process and possibly find a better solution. The repair operator is used to fill the gap created by the destroy operator, thus constructing a new complete route. The detailed application of these operators is as follows:

[0097] The 7 destroy operators are as follows:

[0098] a) Random removal: randomly remove nodes from the current solution, where is a predefined integer, usually greater than 1.

[0099] b) Cluster removal: randomly select a node as the focal node, then remove the node and its nearest nodes.

[0100] c) Nearest customer removal: randomly remove a node from a path, and remove the nearest task nodes in the path.

[0101] d) Path removal: randomly select a truck path, and remove all truck and drone nodes related to it.

[0102] e) Worst cost removal: remove the node that can save the most cost after deleting nodes in the current solution.

[0103] f) Random removal-drone: randomly select a drone, remove all nodes visited by it, and unload the drone from the truck it belongs to.

[0104] g) Worst cost removal-drone: select the drone with the largest travel cost, remove all nodes visited by it, and unload the drone from the truck it belongs to.

[0105] The 9 repair operators are as follows:

[0106] a) Random insertion-truck: randomly select an insertion position that satisfies the warehouse time window constraint, then insert the truck node. If no suitable insertion position is found, create a new truck route.

[0107] b) Nearest insertion-truck: try to insert the unvisited truck node into an existing path, the position is the nearest truck node. If this insertion does not satisfy the time window constraint, consider the second nearest node, and so on. If there is no suitable position, create a new truck route.

[0108] c) Greedy insertion - truck: Compare the cost increase when inserting a truck node at all possible locations (including creating a new route), and choose the insertion location with the smallest cost increase.

[0109] d) New truck route construction: Connect all unvisited truck nodes into a new route, using the same method as in step 1 of the initialization phase. This is only performed if there is a free truck.

[0110] e) Random insertion - drone: Randomly choose an insertion strategy for unvisited drone nodes, choosing from the following two options: (1) create a new drone trip; (2) insert into a feasible location in an existing drone trip.

[0111] f) Nearest insertion - drone: Similar to step 2 of the initialization phase, create a new drone trip for unvisited drone nodes, taking off from the nearest truck node or inserting into the nearest existing drone trip.

[0112] g) Greedy insertion - drone: Choose the insertion location that results in the smallest cost increase, either by inserting the node into an existing route or creating a new drone trip.

[0113] h) Activate new drone - random: Activate a new drone for each type of unvisited drone node, assigning it to the nearest truck to that type of unvisited node. Then, use the new drone to serve the unvisited drone nodes by randomly choosing a feasible way (new trip or insert into existing trip). Inaccessible nodes are handled by existing drones using the same strategy as random insertion - drone.

[0114] i) Activate new drone - nearest: Activate a new drone for each type of unvisited drone node and assign it to the nearest truck to that type of unvisited node. Then, use the new drone to serve the unvisited drone nodes by creating a new trip taking off from the nearest truck node or inserting the node into the nearest existing drone trip. Inaccessible nodes are handled by existing drones using the same strategy as nearest insertion - drone.

[0115] These operations are performed through the alternation of destruction and repair, effectively optimizing the scheduling and path planning of trucks and drones, improving the performance and efficiency of the overall system.

[0116] In the present invention, the destruction and repair operators will typically undergo adaptive weight updates based on their performance and be selected using a roulette wheel approach to ensure that the algorithm can continuously and effectively explore the solution space and find high-quality solutions. The specific way to update the weights of the operators is as follows:

[0117] ,

[0118] in, Represents the weights before and after the operator x is updated, and the parameters Indicates the score update weight coefficient, with a value range of 0-1. represents the number of times the operator x appears in the iteration process, It represents the score of each operator in the iterative process. Specifically, when a new optimal solution is obtained, the score is 10; when a solution that is not optimal but better than the current solution is obtained, the score is 5; when a solution that is worse than the current solution but is selected based on the simulated annealing mechanism is obtained, the score is 1.

[0119] Each destruction operator and repair operation has a selection weight , in each iteration, the probability of the operator being selected is determined according to the following formula :

[0120] ,

[0121] X represents the number of operators.

[0122] (3-3) Taking the total cost as the target value, the simulated annealing acceptance criterion is used to determine whether the newly generated route is accepted and update the local optimal solution.

[0123] Compare the cost of the new solution (route) S' with the total cost of the optimal solution. If the total cost of the new solution S' is smaller, then update the optimal solution and the current solution to the new solution S'; if the total cost of the new solution S' is greater than the cost of the optimal solution but less than the cost of the current solution S, then update the current solution to the new solution S'; if the cost of the new solution S' is greater than the cost of the current solution S, then determine whether to retain the new solution S' according to the following formula.

[0124] ,

[0125] Where, T represents the simulation temperature;

[0126] (3-4) Determine whether the iteration termination condition is met. If so, output the local optimal solution as the optimal route and truck-drone configuration plan. Otherwise, Iteration = Iteration + 1, and return to step (3-2).

[0127] (4) According to the spatial distribution characteristics of different tasks and the impact of drone power consumption on the total cost, the optimal route and optimal drone-truck configuration plan for different mission scenarios are adjusted.

[0128] The specific steps are as follows:

[0129] (4-1) Task space distribution modeling is performed, and two typical task space layouts are constructed, including (1) linear distribution: the task points are distributed in a strip shape, which is suitable for scenarios such as river bank patrol and power transmission line inspection; (2) cluster distribution: the task points are concentrated in a local area, which is suitable for wind turbine monitoring scenarios.

[0130] (4-2) Sensitivity analysis is performed by adjusting parameters, and multiple groups of upper limits of the power of unmanned aerial vehicles are set to simulate the system performance under different endurance capabilities, and the influence of the upper limits on the path selection and the total cost of the system is analyzed; different proportions of trucks performing tasks and unmanned aerial vehicles performing tasks are set to compare the system cost and the vehicle fleet configuration under different task distributions.

[0131] (4-3) Decision suggestions are proposed for specific scenarios, for example, for cluster distribution tasks, it is recommended to deploy more unmanned aerial vehicles for each truck to reduce the truck path length; for linear distribution tasks, it is suggested to use unmanned aerial vehicles with stronger endurance and appropriately increase the number of truck points to improve the coverage.

[0132] The following numerical experiments are performed on the use examples of the present application to verify the feasibility of the model and algorithm, as follows:

[0133] Since the problem is relatively complex, there is no ready-made standard example available for direct use. In order to verify the effectiveness and feasibility of the method proposed in the present application, the present application adopts the strategy of generating test examples based on the existing literature to create test examples, which can simulate real-world task scenarios, thereby comprehensively testing and verifying the vehicle fleet and path joint optimization method of the present application.

[0134] The present application uses 18 examples for experimental verification, and the number of nodes in the scene gradually increases from 8 to 32. When solving each example, the results of the algorithm (ALNS) proposed in the present application are compared with the results of the commercial solver Gurobi, as shown in Table 1, and the visualization of the solution results of the examples is shown in Figure 4 For each example, the ALNS algorithm is executed 10 times using different random seeds. As shown in the third and sixth columns of Table 1, the calculation time of Gurobi increases rapidly with the increase of the problem size (the maximum limit is 48 hours), while the ALNS algorithm in the present application can always solve the problem in a short time. In addition, the difference between the most target values reported in the seventh column is zero, indicating that the ALNS algorithm successfully found the global optimal solution in many instances. In addition, the difference between the average target value and the best target value obtained by the ALNS is always kept below 3%, which demonstrates the robustness and effectiveness of the algorithm in producing high-quality, near-optimal solutions.

[0135] Table 1 Solution results on examples

[0136] Example encoding Gurobi optimal solution Gurobi solution time (s) ALNS optimal solution ALNS average solution ALNS solution time (s) Gap between ALNS optimal solution and Gurobi optimal solution (%) Gap between ALNS average solution and Gurobi optimal solution (%) 8-2-1-2A 156 0.07 156 156 0.03 0.00 0.00 8-2-1-2B 154 0.06 154 154 0.04 0.00 0.00 10-2-1-2A 156 0.26 156 156 0.17 0.00 0.00 10-2-1-2B 174 0.27 174 174 0.19 0.00 0.00 12-2-2-4A 178 32.86 178 178.8 1.54 0.00 0.04 12-2-2-4B 192 17.9 192 192.4 1.1 0.00 0.04 14-2-2-4A 204 12.46 204 204.4 2.28 0.00 0.02 14-2-2-4B 232 20.68 232 233.6 3.74 0.00 0.69 16-2-2-4A 210 178.04 210 210.8 1.81 0.00 0.38 16-2-2-4B 274 207.06 274 274.6 2.74 0.00 0.22 18-2-2-4A 254 2431.97 254 254.6 9.05 0.00 0.24 18-2-2-4B 292 430.82 296 297.2 10.76 1.37 1.78 20-2-2-4A 256 874.25 256 257.4 17.45 0.00 0.55 20-2-2-4B 278 1075.33 278 279.0 24.91 0.00 0.36 24-2-2-6A 382 3118.12 382 386.4 28.73 0.00 1.15 24-2-2-6B 258 7912.33 258 258.6 30.95 0.00 0.23 32-2-2-8A 598 / 598 610.9 61.31 0.00 2.16 32-2-2-8B 554 / 554 569.5 67.25 0.00 2.80

[0137] In summary, for the problem of truck and UAV cooperation to perform heterogeneous inspection tasks, the application successfully constructs a mixed integer programming model that accurately reflects the complexity of the task, which comprehensively considers the heterogeneous requirements of different types of inspection tasks on equipment, the battery endurance of UAVs, the formation of truck and UAV teams, and various constraint conditions such as the time window of the task, can effectively optimize resource allocation and task routes in complex actual environment, and has practical application value. In addition, the application proposes an adaptive large neighborhood search algorithm, forming a solution framework for truck-UAV cooperative path planning. The framework can flexibly cope with different size and type of task demand scenarios, quickly give high-quality solutions, ensure flexible deployment of heterogeneous UAVs, and complete all tasks using a small number of UAVs and trucks, greatly improving the operational flexibility of the system and reducing system costs, thereby providing an efficient solution for the safety inspection industry.

[0138] Embodiment two

[0139] The embodiment of the application further provides a computer program product, such as an app on a mobile phone, a tablet computer, an installation program on a computer, etc. The product includes computer programs / instructions, which, when executed by a processor, implement the method of embodiment one. The code for the computer executable program for executing the operation of the application can be written in one or more programming languages or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" language or similar programming languages. The program code can be executed entirely on a user computer, partially on a user computer, as a separate software package, partially on a user computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, through the Internet using an Internet service provider).

[0140] It should be understood that the above embodiments and descriptions described in the specification are only the principles, main features and advantages of the application, and various changes and improvements can be made to the application without departing from the spirit and scope of the application. These changes and improvements all fall within the scope of the application.

Claims

1. A truck-UAV cooperative path optimization method for heterogeneous inspection tasks, characterized in that, The steps include: (1) The objective function of the truck-UAV collaborative path optimization model is set as: ， wherein, denotes the total cost, denote the truck start-up cost, the unit time travel cost, denote the drone start-up cost, the unit time travel cost, are truck travel route variables, which are equal to 1 when the truck travels from the warehouse to the task node , from the task node to the task node , and from the task node to the end point, respectively, and are equal to 0 otherwise, is a drone-truck assignment variable, which is equal to 1 when the drone is assigned to the truck k, and is equal to 0 otherwise, denote the travel time of the truck and the drone from the node to the node , respectively, is a drone route variable, which is equal to 1 when the drone travels from the task node to the task node , and is equal to 0 otherwise, i is a set of all task nodes except the warehouse and the end point, is a set of all task nodes except the end point, is a set of all task nodes except the start point, is a set of all drones, is a set of all trucks, is a set of all drone trips, are decision variables;​​​​​​​​​ (2) Add routing constraints, heterogeneity constraints on equipment for different types of inspection tasks, and UAV energy consumption and time constraints to the truck-UAV collaborative path optimization model; (3) Solve the truck-UAV collaborative path optimization model to obtain the optimized decision variables, thereby obtaining the optimal route and optimal UAV-truck configuration solution for heterogeneous inspection tasks; Step (3) specifically includes: (3-1) Based on the preset initial truck-drone configuration plan, the initial mission execution route of the truck and drone is constructed using the nearest neighbor insertion method. After the route is constructed, all idle drones are removed and returned to the warehouse; (3-2) Using a roulette wheel strategy to select and apply destruction operators and repair operators to generate a new route, and introduce or retract drones; the destruction operator is used to remove some nodes from the current route, thereby creating an incomplete route, and the repair operator is used to fill the gaps created by the destruction operator, thereby constructing a new complete route; (3-3) Taking the total cost as the target value, use the simulated annealing acceptance criterion to determine whether the newly generated route is accepted and update the local optimal solution; (3-4) Determine whether the iteration termination condition is met. If so, output the local optimal solution as the optimal route and truck-UAV configuration plan. Otherwise, return to step (3-2).

2. The truck-UAV collaborative path optimization method for heterogeneous inspection tasks according to claim 1, characterized in that, The routing constraints include: ， ， ， ， ， wherein is a truck route variable, is a truck-task configuration variable, which is 1 when the task node is visited by the truck and 0 otherwise, , denote the position of the task node j in the truck route, denotes the number of task nodes in , and denotes the drone take-off point variable, which is 1 when the drone takes off from in flight f and 0 otherwise, denotes the drone take-off point variable, which is 1 when the drone is taken in at in flight f and 0 otherwise.

3. The truck-UAV collaborative path optimization method for heterogeneous inspection tasks according to claim 1, characterized in that, The routing constraints also include: , , , , , , , , , , , wherein, are both drone trip variables, respectively taking value 1 when the task node is visited in the drone's trip f, and 0 otherwise, , both represent drone take-off point variables, , represent drone recovery point variables, represent drone route variables, is a truck-task configuration variable, represents the set of all task nodes, represents the maximum number of drones the truck can carry.​ 4. The truck-UAV collaborative path optimization method for heterogeneous inspection tasks according to claim 1, characterized in that, The heterogeneous constraints imposed by different types of inspection tasks on equipment include: , , wherein, is a truck-task configuration variable, is a node-truck inspection variable, when a task node can be inspected by a truck is 1, otherwise 0, is a node-drone inspection variable, when a task node can be inspected by a drone d is 1, otherwise 0, is a drone trip variable.

5. The truck-UAV collaborative path optimization method for heterogeneous inspection tasks according to claim 1, characterized in that, The energy consumption constraints of the UAV include: , , , , , In the formula, , respectively represent the unmanned aerial vehicle The remaining power at the task node j, represent the unmanned aerial vehicle battery capacity, represent the unmanned aerial vehicle take-off point variable, represent the unmanned aerial vehicle recovery point variable, represent the inspection time of the task node , is the unmanned aerial vehicle travel variable.

6. The truck-UAV collaborative path optimization method for heterogeneous inspection tasks according to claim 1, characterized in that, The time constraints include: , , , , , , , , , , , In the formula, , respectively represent the truck departure time from the warehouse, task node , respectively represent the warehouse time window start time, end time, is a truck travel route variable, respectively represent the truck arrival time at the task node , represent the task node inspection time, respectively represent the UAV arrival time at the task node , represent the UAV battery replacement time, represent the UAV takeoff point variable, is a truck-task configuration variable, represent the truck arrival time at the end point, represent the UAV departure time from the task node , is a UAV travel variable.

7. The truck-UAV collaborative path optimization method for heterogeneous inspection tasks according to claim 1, characterized in that, The time constraints also include: , , , , wherein, denotes the warehouse time window end time, denotes the truck arrival time at the task node , denotes the task node inspection time, denotes the drone arrival time at the task node , denotes the drone battery replacement time, , both denote the drone takeoff point variable, is the truck-task configuration variable, denotes the drone arrival time at the end point, denote the drone departure time from the task node , denotes the drone pickup point variable.

8. The truck-UAV collaborative path optimization method for heterogeneous inspection tasks according to claim 1, characterized in that, After step (3), the following steps are also included: (4) According to the spatial distribution characteristics of different tasks and the impact of drone power consumption on the total cost, the optimal route and optimal drone-truck configuration plan for different mission scenarios are adjusted.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images