Long-flight UAV mission planning method, device and computer equipment

Abstract

The present invention relates to a long-flight UAV mission planning method, apparatus, and computer equipment. The method comprises: obtaining mission planning parameters; constructing a long-flight UAV mission planning model; solving the long-flight UAV mission planning model using a pre-set two-layer optimization method based on the mission planning parameters, outputting a current UAV mission planning scheme and a corresponding optimal GCR allocation strategy; the UAV mission planning scheme is used to plan missions for the long-flight UAV; and the optimal GCR allocation strategy is used to allocate tasks to each GCR in a set of GCRs. This method can allocate the most appropriate GCR to each UAV, thereby maximizing overall benefits.

Description

Technical Field

[0001] The present invention relates to the technical field of long-flight UAV mission planning, and in particular to a long-flight UAV mission planning method, device and computer equipment. Background Art

[0002] In recent years, drones, as advanced aircraft, have been widely used in various fields, including reconnaissance and surveillance, communication relay, disaster management, logistics distribution, and remote sensing mapping. Based on their endurance, drones are generally divided into two categories: short-endurance and long-endurance. Short-endurance drones have slower flight speeds and shorter ranges due to payload and energy configuration limitations. In contrast, long-endurance drones (LE-UAVs) have longer flight times and heavier payloads, enabling them to perform more diverse and wide-area missions.

[0003] Currently, long-flight drones already possess a certain degree of intelligence and autonomy, and are able to automatically perform operations such as cruising and obstacle avoidance after setting a route. However, long-flight drones are still unable to fully rely on onboard computers to automatically perform operational tasks. Currently, operators are still required to send instructions through a ground control station to control the drone to perform tasks. This mode is called a "man-in-the-loop" command and control mode. For example, after takeoff, the pilot and image sensor operator at the ground control station jointly operate the drone to complete the mission. In this study, the drone operator is defined as a ground control resource (GCRs).

[0004] The problem of multi-UAV mission planning has been extensively studied. To improve the efficiency and intelligent coordination of multi-UAV missions in complex environments, scholars have proposed a variety of mission planning algorithms, including mathematical programming, heuristic algorithms, swarm intelligence algorithms, and contract network algorithms. However, most current research assumes that UAVs possess full autonomy and do not require GCR control. In practical applications, long-endurance UAVs do not possess full autonomy and still require GCR control during mission execution. Although a few studies have considered the need for GCR control, these studies have overlooked the significant impact of GCR heterogeneity (different mission capabilities) and fatigue on mission performance when planning multi-UAV missions. In fact, literature has shown that GCR capabilities and fatigue have a significant impact on UAV operation. Furthermore, current research assumes that the communication connection between UAVs and GCRs is fixed, that is, one GCR is fixedly connected to one UAV. In reality, a flexible matching mechanism exists between long-endurance UAVs and GCRs, allowing UAVs to select the most appropriate GCR to control their operations based on current mission information.

[0005] How to centrally optimize drone mission planning and GCR allocation, rationally assigning mission sequences to each drone based on each operator's ability and fatigue, and simultaneously assigning the most appropriate GCR to each drone to maximize overall benefits, has become a pressing issue. Summary of the Invention

[0006] Based on this, it is necessary to provide a long-flight UAV mission planning method, device and computer equipment to address the above technical problems.

[0007] A long-flight UAV mission planning method, the method comprising:

[0008] Obtaining mission planning parameters; the mission planning parameters include a fully connected graph of the service customer, a planning time range, the execution time at the customer node, the flight time required from customer node i to customer node j, the number of resources required by the customer node, the amount of resources carried by the drone, the drone's service capability for the customer node, the drone's GCRs set, and the initial capability value of the GCR for the customer node; the fully connected graph includes nodes and edges, where nodes include the customer node and the start and end points of the drone, and edges represent the positional relationships between customers;

[0009] A long-flight UAV mission planning model is constructed; the goal of the long-flight UAV mission planning model is to maximize the mission benefit function; the mission benefit function includes upper-level benefits, lower-level benefits, and UAV flight time cost; the constraints of the long-flight UAV mission planning model include starting point constraints, end point constraints, customer node flow conservation constraints, UAV visit uniqueness constraints, UAV arrival time constraints, unvisited customer node constraints, UAV flight time constraints, UAV resource constraints, coupling constraints between GCR allocation decisions and time scheduling, and GCR uniqueness constraints;

[0010] According to the mission planning parameters, the long-flight UAV mission planning model is solved using a preset two-layer optimization method to output the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy; the UAV mission planning scheme is used to plan missions for long-flight UAVs; the optimal GCR allocation strategy is used to allocate tasks to each GCR in the GCRs set.

[0011] A long-flight UAV mission planning device, comprising:

[0012] A parameter acquisition module is used to obtain mission planning parameters; the mission planning parameters include the fully connected graph of the service customer, the planning time range, the execution time at the customer node, the flight time required from customer node i to node j, the number of resources required by the customer node, the amount of resources carried by the drone, the drone's service capability for the customer node, the drone's GCRs set, and the initial capability value of the GCR for the customer node; the fully connected graph includes nodes and edges, where nodes include the customer node and the start and end points of the drone, and edges represent the positional relationship between customers;

[0013] A model building module is used to build a long-flight UAV mission planning model; the goal of the long-flight UAV mission planning model is to maximize the mission benefit function; the mission benefit function includes upper-level benefits, lower-level benefits, and UAV flight time cost; the constraints of the long-flight UAV mission planning model include starting point constraints, end point constraints, customer node flow conservation constraints, UAV visit uniqueness constraints, UAV arrival time constraints, unvisited customer node constraints, UAV flight time constraints, UAV resource constraints, GCR allocation decision and time scheduling coupling constraints, and GCR uniqueness constraints;

[0014] The result output module is used to solve the long-flight UAV mission planning model according to the mission planning parameters using a preset two-layer optimization method, and output the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy; the UAV mission planning scheme is used to plan the mission of the long-flight UAV; the optimal GCR allocation strategy is used to allocate tasks to each GCR in the GCRs set.

[0015] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0016] Obtaining mission planning parameters; the mission planning parameters include a fully connected graph of the service customer, a planning time range, the execution time at the customer node, the flight time required from customer node i to customer node j, the number of resources required by the customer node, the amount of resources carried by the drone, the drone's service capability for the customer node, the drone's GCRs set, and the initial capability value of the GCR for the customer node; the fully connected graph includes nodes and edges, where nodes include the customer node and the start and end points of the drone, and edges represent the positional relationships between customers;

[0017] A long-flight UAV mission planning model is constructed; the goal of the long-flight UAV mission planning model is to maximize the mission benefit function; the mission benefit function includes upper-level benefits, lower-level benefits, and UAV flight time cost; the constraints of the long-flight UAV mission planning model include starting point constraints, end point constraints, customer node flow conservation constraints, UAV visit uniqueness constraints, UAV arrival time constraints, unvisited customer node constraints, UAV flight time constraints, UAV resource constraints, coupling constraints between GCR allocation decisions and time scheduling, and GCR uniqueness constraints;

[0018] According to the mission planning parameters, the long-flight UAV mission planning model is solved using a preset two-layer optimization method to output the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy; the UAV mission planning scheme is used to plan missions for long-flight UAVs; the optimal GCR allocation strategy is used to allocate tasks to each GCR in the GCRs set.

[0019] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0020] Obtaining mission planning parameters; the mission planning parameters include a fully connected graph of the service customer, a planning time range, the execution time at the customer node, the flight time required from customer node i to customer node j, the number of resources required by the customer node, the amount of resources carried by the drone, the drone's service capability for the customer node, the drone's GCRs set, and the initial capability value of the GCR for the customer node; the fully connected graph includes nodes and edges, where nodes include the customer node and the start and end points of the drone, and edges represent the positional relationships between customers;

[0021] A long-flight UAV mission planning model is constructed; the goal of the long-flight UAV mission planning model is to maximize the mission benefit function; the mission benefit function includes upper-level benefits, lower-level benefits, and UAV flight time cost; the constraints of the long-flight UAV mission planning model include starting point constraints, end point constraints, customer node flow conservation constraints, UAV visit uniqueness constraints, UAV arrival time constraints, unvisited customer node constraints, UAV flight time constraints, UAV resource constraints, coupling constraints between GCR allocation decisions and time scheduling, and GCR uniqueness constraints;

[0022] According to the mission planning parameters, the long-flight UAV mission planning model is solved using a preset two-layer optimization method to output the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy; the UAV mission planning scheme is used to plan missions for long-flight UAVs; the optimal GCR allocation strategy is used to allocate tasks to each GCR in the GCRs set.

[0023] The above-mentioned long-flight UAV mission planning method, device and computer equipment first obtain the current mission planning parameters for the current long-flight UAV mission planning problem, then call the constructed long-flight UAV mission planning model, use the obtained known parameter data to perform two-layer optimization on the long-flight UAV mission planning model, output the optimal planning scheme that maximizes the mission benefit function and satisfies the constraints of the model, that is, obtain the current UAV mission planning scheme, and finally use the UAV mission planning scheme to plan the mission of the long-flight UAV, and use the corresponding optimal GCR allocation strategy to assign tasks to each GCR in the GCRs set, thereby achieving a more effective implementation of reasonably allocating task sequences to each UAV while selecting the most appropriate GCR for each UAV, thereby maximizing the efficiency of "human-machine integration". BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flowchart of a long-flight UAV mission planning method according to an embodiment;

[0025] Figure 2 Schematic diagram of the algorithm framework of a two-layer optimization method in one embodiment;

[0026] Figure 3 A schematic diagram of the structure of a solution to the HG&U IOP problem in one embodiment;

[0027] Figure 4 FIG. 1 is a schematic diagram of a neighborhood structure in an embodiment, wherein: Figure 4 (a) is a schematic diagram of the reinsertion operation. Figure 4 (b) is a schematic diagram of the exchange operation. Figure 4 (c) is a schematic diagram of the reversal operation;

[0028] Figure 5 A schematic diagram of a GCR time plan in one embodiment;

[0029] Figure 6 A schematic diagram of a modification operation flow in one embodiment;

[0030] Figure 7 This is a structural block diagram of a long-flight UAV mission planning device in one embodiment;

[0031] Figure 8 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0032] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0033] Traditional approaches assume that after takeoff, drones and GCSs are paired in a fixed one-to-one or many-to-one relationship, meaning that one or more UAVs are fixedly controlled by a single GCS. However, drones actually possess a degree of autonomous intelligence, enabling them to patrol autonomously without requiring full control. Ground control stations only intervene to execute missions at their target locations. A flexible many-to-many matching mechanism exists between drones and ground stations. Therefore, drones can fully select the most appropriate GCR to control their operations based on current mission information. To better align with reality and maximize the effectiveness of "human-machine integration," it is necessary to address the issue of how to rationally assign mission sequences to each drone while also selecting the most appropriate GCR for each drone.

[0034] Based on the relationship between long-endurance UAVs and GCRs and their practical applications, the present invention proposes the integrated optimization of multi-long endurance UAV mission planning and multi-heterogeneous GCR assignment (HG&U IOP). This problem is formulated as a mixed integer linear programming (MILP) model with the goal of maximizing overall benefits. To effectively solve the MILP, the present invention proposes a bi-level programming algorithm (BPA) based on the hybrid genetic algorithm (HGA) framework. The upper layer of the BPA is task planning, which is used to decide the task sequence for each UAV. Specifically, the present invention designs three initial solution generation methods based on the problem characteristics, and adopts local search-variable neighborhood descent (LS-VND) to improve the quality of the solution. The lower layer of BPA is heterogeneous GCR allocation, which is used to decide and control the GCR of each UAV for each mission. Specifically, the present invention uses a greedy algorithm combined with local search to calculate the optimal GCR allocation strategy for each chromosome in the HGA. Finally, the benefits of the two layers are weighted and summed using the HGA's fitness function, and iterative optimization is performed until convergence conditions are reached. Numerical experiments verify that BPA can effectively solve the HG&U IOP, and the algorithm's solution performance is superior to that of the advanced optimization toolkit CPLEX. In particular, when processing large-scale examples, the solution obtained by the proposed algorithm in 5 seconds is better than the solution obtained by CPLEX in 2 hours.

[0035] In one embodiment, Figure 1 As shown, a long-flight UAV mission planning method is provided, comprising the following steps:

[0036] Step 102: Obtain mission planning parameters. Mission planning parameters include the fully connected graph of the service customer, the planning time range, the execution time at the customer node, the flight time between the two nodes, the number of resources required by the customer node, the amount of resources carried by the drone, the drone's service capability for the customer node, the drone's GCRs set, and the initial capability value of the GCR for the customer node. The fully connected graph includes nodes and edges. Nodes include the customer node and the starting and ending points of the drone. Edges represent the location relationship between customers.

[0037] Assumptions for the HG&U IOP questions include:

[0038] (1) There is only one airport, and all drones depart from the airport and return to the original airport after completing their mission;

[0039] (2) The total duration of the UAV mission is fixed, and the service time and resource requirements of the target point are fixed;

[0040] (3) Ground stations, drones and satellites are visible in real time, and the signals are not interfered with;

[0041] (4) Within a given time, each control task cannot be interrupted once it starts;

[0042] (5) Any GCR can only control one UAV at a time, and a UAV can only be controlled by one GCR at a time.

[0043] These basic assumptions are intended to simplify the complexity of HG&U IOP while retaining its core elements, allowing for greater focus on studying how to optimize the overall effectiveness of multiple UAVs through reasonable control strategies.

[0044] The mission planning parameters obtained in step 102 are explained as follows:

[0045] Fully connected graph G = (N, A), where the node set N = {0, 1, 2, ..., n}, and the edge set A = {(i, j) | i, j∈N, i≠j}. In the node set, node 0 represents the airport where the drone takes off and lands, which is the starting point and end point of the drone. Represents the set of profitable nodes, i.e., customer nodes.

[0046] The planning time range is expressed as T = {0,1,2,...,t max},in Consider a group of identical drones u in If the drone u departs from the airport at any time and serves a specific customer, the flight time of the drone u shall not exceed exist Once the drone starts serving a customer, the process cannot be interrupted. The service time at customer i, that is, the execution time of the task, is t i , the flight time from node i to node j is t ij The number of resources required by customer i is R i , the resources that drone u can carry are Q u The capability of drone u to customer i can be expressed as

[0047] The set of GCRs of the drone is represented by S. In the process of flying to node i, the drone can fly autonomously without GCR control. However, when providing services to customers, the drone needs the control of GCR to perform tasks. Each GCR has different capabilities for each customer. For customer i, the initial capability value of GCR can be expressed as The GCR capability value gradually decreases with the accumulation of working time, and the change function is: The definition is as shown in formula (1)

[0048]

[0049] Where β is a user-defined parameter that represents the rate of decrease in GCR capability. When its value increases, the GCR capability decreases rapidly over time.

[0050] Step 104, construct a long-flight UAV mission planning model; the goal of the long-flight UAV mission planning model is to maximize the mission benefit function; the mission benefit function includes upper-level benefits, lower-level benefits, and UAV flight time cost; the constraints of the long-flight UAV mission planning model include starting point constraints, end point constraints, customer node flow conservation constraints, UAV visit uniqueness constraints, UAV arrival time constraints, unvisited customer node constraints, UAV flight time constraints, UAV resource constraints, GCR allocation decision and time scheduling coupling constraints, and GCR uniqueness constraints.

[0051] The goal of mission planning is to maximize the sum of the benefits of all missions. The benefits of completing a mission depend on the initial value of the customer, the time it takes for the drone to reach the customer, the capabilities of the drone, and the capabilities of the GCR.

[0052]

[0053] Where w1, w2 and w3 are the weights of arrival time, UAV capability and GCR capability respectively. j represents the initial value of the customer. ψ(t) is a monotonically decreasing function of time and can be defined as Equation (3).

[0054]

[0055] Where γ is a user-defined parameter that represents the rate of decrease in the task's benefit. When its value increases, the benefit decreases rapidly over time.

[0056] Traditional drone mission planning primarily focuses on determining the order in which to visit target points and planning the path. Building on this foundation, the allocation of the drone's GCR is also factored into the decision-making process. Furthermore, factors such as the drone's flight duration and payload capacity are considered. Aiming to maximize the benefits of all missions, a long-duration drone mission planning model is constructed.

[0057] Step 106: Based on the mission planning parameters, a pre-set two-layer optimization method is used to solve the long-flight UAV mission planning model, and the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy are output; the UAV mission planning scheme is used to plan the mission of the long-flight UAV; and the optimal GCR allocation strategy is used to allocate tasks to each GCR in the GCRs set.

[0058] Some off-the-shelf solvers (such as CPLEX) can solve small-scale instances, but cannot handle large-scale instances. The present invention adopts a pre-set two-level optimization method, namely a two-level programming algorithm based on a hybrid genetic algorithm framework to solve the long-flight UAV mission planning model.

[0059] In the above-mentioned long-flight UAV mission planning method, the current mission planning parameters are first obtained for the long-flight UAV mission planning problem that needs to be carried out currently, and then the constructed long-flight UAV mission planning model is called, and the obtained known parameter data are used to perform a two-layer optimization on the long-flight UAV mission planning model, and the output is the optimal planning scheme that maximizes the task benefit function and satisfies the constraints of the model, that is, the current UAV mission planning scheme is obtained, and finally the UAV mission planning scheme is used to plan the mission of the long-flight UAV, and the corresponding optimal GCR allocation strategy is used to assign tasks to each GCR in the GCRs set, thereby achieving a more effective implementation of reasonably allocating task sequences to each UAV while selecting the most appropriate GCR for each UAV, thereby maximizing the efficiency of "human-machine integration".

[0060] In one embodiment, the task benefit function is:

[0061]

[0062] The constraints are:

[0063]

[0064]

[0065] The relevant parameter definitions are shown in the following table:

[0066]

[0067] The objective function (4) maximizes the overall benefit of the system, where is the flight time cost of the UAV from takeoff to landing. Constraint (5) defines ψ(t′ j ) parameter t′ j The calculation method of constraint condition (6) is defined as Parameter t j ″ calculation method. Constraints (7)-(8) represent the starting point constraint and the end point constraint, respectively, ensuring that each aircraft departs from the airport and eventually returns to the airport. Constraint (9) is the flow conservation constraint of the customer node, which is the flow conservation of each customer node. Constraint (10) is the drone access uniqueness constraint, which means ensuring that one and only one drone passes through each customer. Approximately (11)-(12) are drone arrival time constraints, which combine routing decisions with time scheduling. Constraint (11) ensures that if x 0,j,u = 1, then the arrival time of the first customer j that UAV u passes from airport 0 must be equal to or greater than the take-off time plus the flight time from airport 0 to customer j. Constraint (12) ensures that if x 0,j,u =1, that is, after leaving customer j, drone u directly visits customer i. Then the time for drone u to arrive at j must be equal to or greater than the time it arrives at i plus the service time for customer i and the flight time between customer i and customer j. Constraint (13) is the unvisited customer node constraint, which represents the virtual node j variable that has not been visited by drone u. is zero. Constraint (14) is the UAV flight time constraint, which indicates the flight time limit and ensures that each UAV can return to the airport. Constraint (15) is the UAV resource constraint, which indicates that each UAV has enough resources to meet the needs of the customer nodes it traverses. Constraint (16) represents the coupling constraint of GCR allocation decision and time scheduling, where the time when GCR arrives at customer i when UAV u is equal to the time when UAV u starts to serve customer i. Constraints (17) and (18) are GCR uniqueness constraints. Constraint (17) indicates that there is only one GCR controlling the UAV to serve customer i. Constraint (18) ensures that the GCR can only control one UAV at the same time. Constraints (19)-(21) define the domain of the decision variables.

[0068] It should be noted that the objective function (4) is nonlinear because it contains fractional and exponential terms. Therefore, we use the pre-calculation and lookup table method to linearize it. Since we use the time index modeling method, the time is discrete, that is, T = {1, 2, ..., t max Therefore, we can pre-calculate the values ​​of the fractional term and the exponential term at each integer time point and store them. In the model, when we need to use a specific time t, we can directly call the pre-calculated values.

[0069]

[0070] The construction of the exponential term and fractional term lookup tables is shown in Equations (23) and (24), respectively. Then, when building the model, we directly use the pre-calculated values ​​in the lookup tables to calculate and, as shown in Equations (25) and (26). Finally, the objective function is linearized, as shown in Equation (27).

[0071] In one embodiment, Figure 2 The algorithm framework diagram of the two-layer optimization method shown in the figure uses a preset two-layer optimization method to solve the long-flight UAV mission planning model, and outputs the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy, including: using the upper-layer algorithm to perform the upper-layer task planning to obtain the initial UAV mission planning scheme; using the lower-layer algorithm to perform the lower-layer GCR allocation to obtain the optimal GCR allocation strategy under the initial UAV mission planning scheme; calculating the task benefit function value under the initial UAV mission planning scheme and the corresponding optimal GCR allocation strategy to obtain the current task benefit; iteratively updating the UAV mission planning scheme and the corresponding task benefit based on the preset hybrid HGA algorithm and the initial UAV mission planning scheme until the iteration stop condition is met, stopping the iteration, and outputting the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy.

[0072] In this embodiment, the structural diagram of the HG&U IOP problem solution is as follows: Figure 3 As shown, Ind = {Total_reward, Reward1, Reward2, Penalty Cost, S_work, Chromosome}, where Total_reward = Reward1 + Reward2 - Cost, indicating that the target value in the solution is the sum of the upper layer reward value Reward1 and the lower layer reward value Reward2 minus the weighted track cost Cost. The calculation formula of Reward1 is obtained by formula (28); the calculation formula of Reward2 is obtained by formula (29); Obtaining S_work is the task sequence that each GCR needs to control. A chromosome is a set of multiple sub-chromosomes. Chromosome = (subchromosome1, subchromosome2, ..., subchromosome |U| ), each sub-chromosome represents a UAV mission planning scheme.

[0073] The task sequence in the sub-chromosome is the order list of visiting customers, which is encoded in integers, where positive integers represent target points and "0" represents drone airports. sub_demand is the demand of all customers in each drone's planned route, and flight_duration is the flight time of each drone from takeoff to landing. Represents the time series of each drone’s take-off time and arrival time at each node.

[0074] In one embodiment, an upper-layer algorithm is used to execute upper-layer task planning to obtain an initial UAV task planning scheme, including: using a pre-set initial solution generation strategy to generate an initial solution that maximizes the upper-layer benefit; the initial solution includes a UAV task sequence corresponding to each UAV; the UAV task sequence includes a time series corresponding to the time when the UAV arrives at the starting point, each customer node and the end point, the UAV planned route, the demand of each customer node on the UAV planned route and the total flight time of the UAV; the initial solution generation strategy includes a randomized nearest neighbor method, a randomized average distribution method and a distributed nearest neighbor method; the neighborhood structure of each UAV task sequence is changed, and a local search is performed to find the result with the maximum upper-layer benefit to obtain an optimized initial solution; a pre-designed crossover operator, a recombination operator and a mutation operator are used to iteratively perform crossover operations, recombination operations and mutation operations on the optimized initial solution until the conditions for stopping the iteration are met, and the iteration is stopped to obtain the initial UAV task planning scheme.

[0075] In this embodiment, the upper layer reward 1 is:

[0076]

[0077] Where t′ j Derived from time_series, other parameter information is consistent with the MILP model, x iju From subchromosome u The task sequence in is decoded.

[0078] The quality of the initial solution is crucial to the solution results and convergence speed of BPA. To improve the quality and richness of the initial population solution, the initial solution generation strategy designed by this invention includes the following three strategies based on the characteristics of the problem.

[0079] (1) Randomized Nearest Neighbor: The nearest neighbor method is a common method in virtual reality (VRP) problems. Based on this method, the present invention makes some improvements. First, a point A with the largest reward1 value from the origin is selected. Then, a greedy strategy is used with a 90% probability to select the point with the largest reward1 value starting from point A. A point is randomly selected with a 10% probability, and so on. When the range and resources are insufficient to support the drone to reach the next node, it returns to the end point.

[0080] (2) Randomized average generation: Unlike traditional drone mission planning problems, the goal in emergency or disaster relief is often to complete all tasks in the shortest possible time. Therefore, the present invention aims to evenly allocate the goals of dispatchable drones, aiming to complete all tasks as quickly as possible. This is accomplished by the following steps:

[0081] Step 1: Disrupt the task sequence.

[0082] Step 2: Assign targets to drones u in the set U one by one in order until the last u is also assigned a target, and the selected target is deleted from the sequence.

[0083] Step 3: Assign the unassigned targets starting from the first u in U and repeat step 2 until the assignment is completed.

[0084] Step 4: Output the mission sequence of each drone.

[0085] (3) Distributed nearest neighbor: In order to evenly distribute tasks to each drone, so that tasks are evenly distributed among each drone and the overall task execution efficiency is accelerated, the point with the largest reward1 value starting from the current node is selected for each drone in order. The specific steps are as follows:

[0086] Step 1: According to the order in the set U, UAV u selects targets in the mission sequence using the Nearest Neighbor method, and the selected targets are deleted from the sequence.

[0087] Step 2: Loop step 1 until the task sequence is empty.

[0088] Step 3: Output the mission sequence of each drone.

[0089] Then, the LS-VND acceleration strategy is used to systematically change the neighborhood structure set of the current solution during the search process to expand the search range and find a better solution. For each UAV's mission sequence, the LS-VND algorithm is executed to search the neighborhood. Local search is a sequential search algorithm, such as Figure 4 The neighborhood structure diagram shown in Figure 2 is used to search for three basic neighborhood structures: reinsertion, which is to relocate the target node from one position in the task sequence to another, such as Figure 4 As shown in (a), the neighborhood set in this structure is defined as Nh1; two target nodes are exchanged from two positions in the task sequence, as shown in Figure 4 As shown in (b), the neighborhood set in this structure is defined as Nh2; the nodes in a part of the reverse task sequence, such as Figure 4 As shown in (c), the neighborhood set in this structure is defined as Nh3.

[0090] The specific algorithm is shown in the following table:

[0091]

[0092] For each new task sequence generated by LS-VND, Reward1 is used to evaluate its quality. Furthermore, the task sequence is updated based on Reward1. From an algorithmic perspective, the adaptability of LS-VND aims to enhance the intensification and diversification of HGA. The core idea of ​​the LS-VND algorithm is to systematically change the neighborhood structure of the current solution, expanding the search range. Then, a local search algorithm is used to find the current optimal solution, resulting in an optimized initial solution.

[0093] Furthermore, according to the characteristics of the problem, the present invention designs a series of crossover, recombination and mutation operators to improve the quality of the solution and the efficiency of the algorithm:

[0094] Subroute exchange crossover (SEC): Randomly select a subroute from the parent chromosome P2 and randomly select fragment frag2 from the subroute. Then, eliminate the elements of frag2 from the parent chromosome P1. Randomly insert fragment frag2 into P1, generating the daughter chromosome C1. The same method is used to generate the daughter chromosome C2.

[0095] Subroute single point crossover (SSPC): A subroute is randomly selected from the parent chromosome P2. A fragment frag2 is randomly selected from the subroute. The elements of frag2 from the parent chromosome P1 are then eliminated. The elements of frag2 are then randomly inserted into P1 one by one, generating the daughter chromosome C1. The same method is used to generate the daughter chromosome C2.

[0096] After crossover, the newly formed offspring chromosomes are subjected to further crossover operations, called recombination optimization operators (ROO). These operations are repeated to find better offspring.

[0097] For the recombination optimization operator ROO of the offspring chromosome C, we design it as follows:

[0098]

[0099] To improve the exploration capability of the genetic algorithm, we designed a simple random mutation: randomly select a subroute from the chromosome, randomly select a point in the subroute, and then randomly select another subroute Sub2 and insert this point into a random position in the subroute Sub2.

[0100] In one embodiment, a lower-layer algorithm is used to perform lower-layer GCR allocation to obtain the optimal GCR allocation strategy under the initial UAV task planning scheme, including: using a greedy algorithm to calculate the GCR initial allocation scheme that maximizes the lower-layer benefit under the initial UAV task planning scheme; the GCR initial allocation scheme includes the UAV planning tasks that each GCR is responsible for; and using a local search method based on an exchange neighborhood structure to optimize the planning tasks included in the conflict set in the GCR initial allocation scheme to obtain the optimal GCR allocation strategy.

[0101] After the upper-level planning decisions are made, a mission plan is obtained for each UAV. However, in real-world applications, it is also necessary to consider the proper allocation of GCRs for these control tasks. Therefore, the lower level of BPA performs heterogeneous GCR allocation. In this layer, the lower-level reward Reward2 is:

[0102]

[0103] Where t′ j Derived from time_series, other parameter information is consistent with the MILP model.

[0104] In one embodiment, a greedy algorithm is used to calculate an initial GCR allocation plan that maximizes the lower-level benefit under the initial UAV task planning plan, including: obtaining a time series in each UAV task sequence in the initial UAV task planning plan, sorting the planning tasks involved in the time series in ascending order according to the start time, and obtaining a task set; traversing the planning tasks in the task set in turn, calculating the benefit value corresponding to the current planning task and each GCR in the GCRs set, and allocating an idle GCR to the current planning task according to the benefit value, until all planning tasks are traversed, generating a work schedule for each GCR, and obtaining the GCR initial allocation plan.

[0105] In this example, based on the upper-level task planning scheme, we first use a greedy algorithm to obtain an initial allocation plan for heterogeneous GCRs. The main idea of ​​this method is to sort all tasks in chronological order and prioritize the GCR with the highest benefit for the task that started first. The specific steps are as follows:

[0106] 1) Sort all tasks in ascending order according to their start time based on time_series to generate a task set G. The GCR set is S, and S_work is initialized. For the first task g in G, the following assignment procedure is used:

[0107] a) Calculate the benefit of all GCRs and the current task g. Sort the GCRs by benefit to get SL.

[0108] b) Assign the GCR s with the largest benefit to task g. If this GCR is free, add the assignment result to S_work[s]. If the GCR is occupied, the previous task g assigned to s is pre If it has not been completed when task g starts, the next GCR in SL is selected until the allocation is successful.

[0109] c) Select the next task to assign.

[0110] 2) The assignment process is iterated until all tasks are assigned GCRs, and finally a work schedule for GCRs is generated.

[0111] The algorithm is shown in the table:

[0112]

[0113] After obtaining the initial allocation scheme for heterogeneous GCRs using the greedy algorithm, we use a local search method based on the exchange domain structure Nh2 to improve the quality of the solution. The specific steps are as follows:

[0114] 1) Starting at t = 0, we iterate over all tasks and define the set of tasks that overlap in time as a conflict set (p). As shown in the figure, tasks {3, 5, 7} form one conflict set, and tasks {2, 4, 6, 8} form another conflict set. Furthermore, the same drone's task can appear only once in a set. Finally, we obtain a list of all conflict sets (sp).

[0115] 2) For each subset p in sp, use the greedy iterative method to search the domain, constantly exchanging their GCRs until no better solution is found.

[0116] 3) Update S_work based on the local search results.

[0117] The algorithm is shown in the table:

[0118]

[0119] In one embodiment, the UAV mission planning scheme and the corresponding mission benefits are iteratively updated based on a preset hybrid HGA algorithm and an initial UAV mission planning scheme until the iteration stop condition is met, the iteration is stopped, and the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy are output. The method includes: in each iteration, the elite offspring and the remaining offspring are divided according to the mission benefits of each offspring, the elite offspring are retained, and crossover, recombination, mutation, and repair operations are performed on the remaining offspring. The above process is iterated until the iteration stop condition is met, the iteration is stopped, and the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy are output.

[0120] In this embodiment, the fitness function in the hybrid genetic algorithm (HGA) is used to evaluate the performance of the solution or chromosome. It plays a key role in the search process of the genetic algorithm and the quality of the final result. The main task of the fitness function is to provide a quantitative score to indicate the quality of the chromosome, thereby guiding the algorithm to evolve towards a better solution. In HGA, fitness mainly consists of two parts: one is the benefit, namely Reward1 and Reward2 in the solution; the other is the penalty value Penalty for the drone violating the constraint, and the track cost Cost. The fitness function can be expressed as:

[0121] f(c)=Reward1+Reward2-Cost-Penalty

[0122] Penalty(c) is calculated as follows:

[0123]

[0124] Among them, U cis the number of drones in this chromosome, d u represents the resource requirement of the UAV u task sequence in this chromosome, Q u is the rated capacity of the UAV, t u is the flight time of the drone u, is the rated flight time of the UAV, i is the current iteration number, and I is the total iteration number.

[0125] The calculation formula of coefficient α is as follows:

[0126]

[0127] Here, b is the total reward value (Total_reward) of the optimal solution in the current population, m is three times the sum of all customers' initial values, and ε is a small positive number to ensure the denominator is non-zero. As iterations proceed, the weight of the penalty term, α, gradually increases. This means that as the algorithm searches for a solution that satisfies all constraints, it gradually increases the penalty for violations. This mechanism helps the algorithm find a solution that satisfies all constraints as quickly as possible.

[0128] In this embodiment, based on the characteristics of the problem, the present invention designs customized crossover, recombination, and mutation operations, as well as a repair operator. Furthermore, an elite retention strategy and a simulated annealing mechanism are introduced to determine the probability of accepting differential solutions in the offspring. These mechanisms not only accelerate the convergence of the algorithm but also enhance its exploration capabilities. The crossover, recombination, and mutation operations have been described above and will not be repeated here.

[0129] Elite retention involves retaining a subset of chromosomes with the highest fitness in each generation, ensuring that the algorithm doesn't lose previously discovered solutions. This strategy is a common technique in genetic algorithms and helps the algorithm converge quickly. The remaining chromosomes in the population undergo normal reproduction, using the same binary tournament method to select two parent chromosomes. These chromosomes then undergo crossover, recombination, mutation, and repair to produce two daughter chromosomes. The normal reproduction process is limited by the crossover rate. This process aims to introduce new gene combinations into the population, helping the algorithm avoid premature convergence to local optima.

[0130] The repair operator plays a vital role in the hybrid genetic algorithm (HGA) to ensure the diversity and quality of the population. Its task is to adjust those solutions that do not meet the constraints (i.e., chromosomes that violate the constraints) so that they meet the actual constraints of the problem. Figure 6As shown in the figure, for the HG&U IOP problem, the main goal of the route repair operator is to ensure that each drone's route meets the constraints in terms of demand and flight time. For solutions that violate the constraints, the repair operator restores the feasibility of the chromosome by selecting the routes with the largest demand and flight time and appropriately redistributing nodes between them. This redistribution is accomplished through a series of insertion and deletion operations, as shown in the figure. For routes that violate the flight time constraint, the repair operator removes the last customer in the route and assigns it to the drone with the shortest flight time; for routes that violate the capacity constraint, the customer is assigned to the drone with the smallest demand.

[0131] Furthermore, the present invention utilizes the concept of the "temperature" parameter in the simulated annealing algorithm to allow some individuals with lower fitness to remain in the population, maintain population diversity, and avoid premature convergence. The higher the temperature, the greater the probability of accepting a poor solution. As the temperature gradually decreases, the probability of accepting a poor solution decreases, the algorithm gradually stabilizes, and is more focused on finding the optimal or near-optimal solution. The probability of accepting a poor solution P is accept The settings are as follows:

[0132]

[0133] By skillfully balancing convergence and diversity, the hybrid genetic algorithm designed in the present invention can efficiently solve task planning problems.

[0134] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0135] In one embodiment, Figure 7 As shown, a long-flight UAV mission planning device is provided, comprising:

[0136] Parameter acquisition module 702 is used to obtain mission planning parameters; mission planning parameters include the fully connected graph of the service client, the planning time range, the execution time at the client node, the flight time between nodes, the number of resources required by the client node, the amount of resources carried by the UAV, the service capability of the UAV for the client node, the set of GCRs of the UAV, and the initial capability value of the GCR for the client node; the fully connected graph includes nodes and edges, where nodes include the client node and the starting and ending points of the UAV, and edges represent the location relationship between clients;

[0137] Model construction module 704 is used to construct a long-flight UAV mission planning model; the goal of the long-flight UAV mission planning model is to maximize the mission benefit function; the mission benefit function includes upper-level benefits, lower-level benefits, and UAV flight time cost; the constraints of the long-flight UAV mission planning model include starting point constraints, end point constraints, customer node flow conservation constraints, UAV visit uniqueness constraints, UAV arrival time constraints, unvisited customer node constraints, UAV flight time constraints, UAV resource constraints, GCR allocation decision and time scheduling coupling constraints, and GCR uniqueness constraints;

[0138] The result output module 706 is used to solve the long-flight UAV mission planning model based on the mission planning parameters using a pre-set two-layer optimization method, and output the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy; the UAV mission planning scheme is used to plan the mission of the long-flight UAV; the optimal GCR allocation strategy is used to allocate tasks to each GCR in the GCRs set.

[0139] The specific definition of the long-flight UAV mission planning device can be found in the definition of the long-flight UAV mission planning method above, and will not be repeated here. The various modules in the above-mentioned long-flight UAV mission planning device can be implemented in whole or in part through software, hardware, or a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules.

[0140] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 8As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a long-flight UAV mission planning method is implemented. The display screen of the computer device can be a liquid crystal display or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0141] Those skilled in the art will understand that Figure 8 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0142] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps of the method in the above embodiment when executing the computer program.

[0143] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method in the above embodiment are implemented.

[0144] Those skilled in the art will appreciate that all or part of the processes in the above-described embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described embodiments. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAM bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0145] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0146] The above-described embodiments merely illustrate several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that a person skilled in the art would be able to make numerous modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A long-flight UAV mission planning method, characterized in that: The method comprises: Obtaining mission planning parameters; the mission planning parameters include a fully connected graph of the service customer, a planning time range, the execution time at the customer node, the flight time between nodes, the number of resources required by the customer node, the amount of resources carried by the drone, the drone's service capability for the customer node, the drone's GCRs set, and the initial capability value of the GCR for the customer node; the fully connected graph includes nodes and edges, where nodes include the customer node and the start and end points of the drone, and edges represent the location relationships between customers; A long-flight UAV mission planning model is constructed; the goal of the long-flight UAV mission planning model is to maximize the mission benefit function; the mission benefit function includes upper-level benefits, lower-level benefits, and UAV flight time cost; the constraints of the long-flight UAV mission planning model include starting point constraints, end point constraints, customer node flow conservation constraints, UAV visit uniqueness constraints, UAV arrival time constraints, unvisited customer node constraints, UAV flight time constraints, UAV resource constraints, coupling constraints between GCR allocation decisions and time scheduling, and GCR uniqueness constraints; According to the mission planning parameters, a pre-set two-layer optimization method is used to solve the long-flight UAV mission planning model, and a current UAV mission planning scheme and a corresponding optimal GCR allocation strategy are output; the UAV mission planning scheme is used to plan the mission of the long-flight UAV; and the optimal GCR allocation strategy is used to allocate tasks to each GCR in the GCRs set; The task benefit function is: in, For upper-level benefits, For the benefit of the lower class, is the change function, is the flight time cost of the UAV from takeoff to landing, The initial value for the customer, is the number of nodes, is the weight of the arrival time, is a monotonically decreasing function of time, is the weight of the UAV capability, Assemble for drones, is the drone serial number, For drones To customers ability, For drones From a node To another node When , the variable is equal to 1, otherwise it is equal to 0. For the GCR collection, For GCR To customers service capabilities, is the weight of GCR capability.

2. The method according to claim 1, characterized in that The method of solving the long-flight UAV mission planning model using a preset two-layer optimization method and outputting the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy includes: The upper-level algorithm is used to execute the upper-level task planning and obtain the initial UAV task planning scheme; The lower-level algorithm is used to perform the lower-level GCR allocation to obtain the optimal GCR allocation strategy under the initial UAV mission planning scheme; Calculate the mission benefit function value under the initial UAV mission planning scheme and the corresponding optimal GCR allocation strategy to obtain the current mission benefit; Based on the preset hybrid HGA algorithm and the initial UAV task planning scheme, the UAV task planning scheme and the corresponding task benefit are iteratively updated until the iteration stop condition is met, the iteration is stopped, and the current UAV task planning scheme and the corresponding optimal GCR allocation strategy are output.

3. The method according to claim 2, characterized in that The upper-level algorithm is used to execute the upper-level task planning to obtain the initial UAV task planning scheme, which includes: A pre-set initial solution generation strategy is used to generate an initial solution that maximizes the upper layer benefit; the initial solution includes a drone mission sequence corresponding to each drone; the drone mission sequence includes a time series corresponding to the time when the drone arrives at the starting point, each customer node, and the end point, the drone's planned route, the demand of each customer node on the drone's planned route, and the total drone flight time; the initial solution generation strategy includes a randomized nearest neighbor method, a randomized average allocation method, and a distributed nearest neighbor method; Change the neighborhood structure of each UAV mission sequence and perform local search to find the result with the maximum upper-level benefit to obtain the optimized initial solution; The pre-designed crossover operator, recombination operator and mutation operator are used to iteratively perform crossover operation, recombination operation and mutation operation on the optimized initial solution until the conditions for stopping the iteration are met. The iteration is stopped and the initial UAV mission planning scheme is obtained.

4. The method according to claim 2, characterized in that The lower-level algorithm is used to perform the lower-level GCR allocation, and the optimal GCR allocation strategy under the initial UAV mission planning scheme is obtained, including: A greedy algorithm is used to calculate an initial GCR allocation plan that maximizes the lower-level benefit under the initial UAV task planning plan; the initial GCR allocation plan includes the UAV planning tasks that each GCR is responsible for; A local search method based on an exchange neighborhood structure is used to optimize the planning tasks contained in the conflict set in the GCR initial allocation scheme to obtain the optimal GCR allocation strategy.

5. The method according to claim 4, characterized in that The greedy algorithm is used to calculate the GCR initial allocation scheme that maximizes the lower-level benefit under the initial UAV mission planning scheme, including: Obtaining a time sequence in each UAV task sequence in the initial UAV task planning scheme, and sorting the planning tasks involved in the time sequence in ascending order according to the start time to obtain a task set; The planned tasks in the task set are traversed in sequence, and the benefit values ​​corresponding to the current planned task and each GCR in the GCRs set are calculated. The idle GCRs are allocated to the current planned task according to the benefit values. After all planned tasks are traversed, a work schedule for each GCR is generated, and an initial GCR allocation plan is obtained.

6. The method according to claim 1, characterized in that Based on the preset hybrid HGA algorithm and the initial UAV mission planning scheme, the UAV mission planning scheme and the corresponding mission benefits are iteratively updated until the iteration stop condition is met, and the iteration is stopped, and the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy are output, including: In each iteration, the elite offspring and the remaining offspring are divided according to the task benefits of each offspring. The elite offspring are retained, and crossover, recombination, mutation, and repair operations are performed on the remaining offspring. The above process is iterated until the conditions for stopping the iteration are met. The iteration is stopped and the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy are output.

7. A long-flight UAV mission planning device applied to the method according to any one of claims 1 to 6, characterized in that: The device comprises: A parameter acquisition module is used to obtain mission planning parameters; the mission planning parameters include the fully connected graph of the service customer, the planning time range, the execution time at the customer node, the flight time between nodes, the number of resources required by the customer node, the amount of resources carried by the drone, the drone's service capability for the customer node, the drone's GCRs set, and the initial capability value of the GCR for the customer node; the fully connected graph includes nodes and edges, where nodes include the customer node and the start and end points of the drone, and edges represent the positional relationship between customers; A model building module is used to build a long-flight UAV mission planning model; the goal of the long-flight UAV mission planning model is to maximize the mission benefit function; the mission benefit function includes upper-level benefits, lower-level benefits, and UAV flight time cost; the constraints of the long-flight UAV mission planning model include starting point constraints, end point constraints, customer node flow conservation constraints, UAV visit uniqueness constraints, UAV arrival time constraints, unvisited customer node constraints, UAV flight time constraints, UAV resource constraints, GCR allocation decision and time scheduling coupling constraints, and GCR uniqueness constraints; The result output module is used to solve the long-flight UAV mission planning model according to the mission planning parameters using a preset two-layer optimization method, and output the current UAV mission planning scheme and the corresponding optimal GCR allocation strategy; the UAV mission planning scheme is used to plan the mission of the long-flight UAV; the optimal GCR allocation strategy is used to allocate tasks to each GCR in the GCRs set.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images