Unmanned aerial vehicle flying joint scheduling method based on improved genetic algorithm in complex environment

Abstract

The invention discloses an unmanned aerial vehicle flying joint scheduling method based on an improved genetic algorithm in a complex environment, and the method comprises the following steps: (1) constructing an unmanned aerial vehicle hangar scheduling model, and detecting space conflicts in a warehouse-out process of an unmanned aerial vehicle in real time in a mode of combining dynamic collision detection with linear interpolation; (2) performing knowledge-based improvement on the genetic algorithm, and introducing a dynamic variation rate and a priority selection strategy to improve optimization efficiency and solution quality; (3) establishing a deck function vehicle path scheduling model, and aiming at minimizing the waiting time of the unmanned aerial vehicle in the processes of refueling, missile loading and traction; (4) through a time synchronization means, a hangar stage and a deck stage are connected into a whole, it is verified that the method can remarkably shorten the total unmanned aerial vehicle flying time in a complex environment, and efficient and stable scheduling performance can still be kept under the requirements of multiple unmanned aerial vehicles; (5) verifying the effect of the joint scheduling mode through a simulation experiment; according to the method, the scheduling time is remarkably shortened, and the quality and stability of a scheduling solution are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of carrier-based UAV scheduling optimization, and in particular to a UAV take-off joint scheduling method based on an improved genetic algorithm in a complex environment. Background Art

[0002] The scheduling of carrier-based UAVs is an important research direction in military applications, involving path planning and resource allocation in multiple tasks and scenarios. However, existing research mostly focuses on the scheduling optimization of specific scenarios, lacking overall optimization considerations for the linkage scheduling of the hangar and the deck. In addition, existing algorithms have problems such as slow convergence speed and low resource utilization rate in complex environments, and it is difficult to meet the actual combat requirements. Therefore, it is of great significance to develop an efficient joint scheduling method suitable for complex environments. Summary of the Invention

[0003] Object of the Invention: The object of the present invention is to provide a UAV take-off joint scheduling method based on an improved genetic algorithm, which optimizes the whole process of UAVs from leaving the hangar to taking off on the deck and solves the problems existing in the background art.

[0004] Technical Solution: A UAV take-off joint scheduling method based on an improved genetic algorithm in a complex environment according to the present invention includes the following steps:

[0005] (1) In the hangar stage, a UAV hangar scheduling model is constructed, and a method combining dynamic collision detection and linear interpolation is used to detect spatial conflicts in real time during the process of UAVs leaving the hangar; dynamic collision detection updates the positions of UAVs moment by moment to determine whether the threat circles of any two UAVs overlap. When the following conditions are met, it is determined as a conflict:

[0006] |O1O2|>R1+R2+d

[0007] Where: O1 and O2 respectively represent the center coordinates of different UAVs; R1 and R2 are the corresponding safety boundary radii of the two, and d is the redundant safety distance;

[0008] (2) Based on the improved genetic algorithm, the global optimization of the UAV scheduling sequence is carried out. The improved genetic algorithm includes the following steps:

[0009] (2a) Population initialization: A random generation and knowledge-assisted hybrid strategy is adopted, 70% of the individuals are randomly arranged, and 30% of the individuals generate the initial sequence based on the principle of proximity to the tractor and the elevator;

[0010] (2b) Dynamic mutation rate control: The mutation rate increases linearly with the number of iterations, and the formula is:

[0011]

[0012] Where, P mutationRepresents the current mutation rate, \(P\). initial and \(P\). final are the initial and final mutation rates respectively, call_count is the number of calls to the mutation function, and max_calls is the assumed maximum number of calls;

[0013] (2c) Selection operation: Adopt an elitist retention and tournament combination strategy, and assign higher selection weights to individuals close to the tractor and the elevator;

[0014] (3) In the deck stage, establish a functional vehicle path scheduling model with the goal of minimizing the waiting time of the refueling vehicle and the ammunition-loading vehicle, subject to the following constraints: Refueling and ammunition loading are mutually exclusive processes, and the vehicle arriving later needs to wait; The moving speed of the functional vehicle is fixed at 30 unit distances per minute, the refueling time is 3 minutes, and the ammunition-loading time is 5 minutes; The objective function is:

[0015]

[0016] where: \(W\). oil,i is the waiting time of the refueling vehicle for the \(i\)-th UAV. \(W\). bomb,i is the waiting time of the ammunition-loading vehicle for the \(i\)-th UAV;

[0017] (4) Connect the hangar and deck stages through a time synchronization mechanism, and record the time when the UAV arrives at the deck parking position in real time. If the deck functional vehicle detects that the UAV has not arrived, it enters the waiting state to achieve coordinated resource scheduling in the two stages.

[0018] Furthermore, the hangar scheduling model needs to satisfy the following constraint conditions:

[0019] (a) At the same time, the tractor or elevator serves only one UAV;

[0020] (b) The tractor processes tasks in a fixed order, and skipping or parallel operations are prohibited;

[0021] (c) The processing capabilities of the tractor and elevator are limited by the maximum processing quantities \(M\) and \(M\). l , satisfying:

[0022]

[0023] where, \(x\). ij represents that the \(i\)-th UAV is processed by the \(j\)-th tractor, and \(y\). jk represents that the \(i\)-th UAV is processed by the \(k\)-th elevator, \(M\). t and \(M\). l are the maximum processing quantities of the tractor and elevator respectively.

[0024] Furthermore, the fitness function of the improved genetic algorithm is:

[0025]

[0026] Among them, the total scheduling time is the sum of the completion times of all UAVs from the hangar to the deck parking position.

[0027] Furthermore, in the deck scheduling model, the UAV status is defined by the following variables:

[0028] AirsStatus[i] ∈ {-1, 0, 1}, where: -1 means refueling, 0 means idle and available to execute new tasks, 1 means carrying ammunition;

[0029] AirsNeedDrags[i] ∈ {0, 0.5, 1.0}, where: 0 means the guarantee is not completed, 0.5 means the guarantee is completed, 1.0 means all guarantees are completed.

[0030] Furthermore, the time synchronization mechanism specifically includes: when the UAV completes the hangar towing and arrives at the deck parking position, update its arrival time T arrival ; the deck functional vehicle starts to perform refueling or ammunition loading tasks after T arrival ; if the functional vehicle arrives early, it will wait until T arrival .

[0031] Furthermore, in dynamic collision detection, the UAV path planning adopts the linear interpolation method, updates the position every 0.1 second, and calculates the overlapping state of the threat circles.

[0032] Furthermore, the deck scheduling model also includes the takeoff position management. The takeoff position status is dynamically marked by the array `Fly_status`, and only one UAV is allowed to use the takeoff position at the same time.

[0033] Furthermore, for the crossover operation of the improved genetic algorithm, the multi-point crossover strategy is adopted. With a 70% probability, crossover occurs at a fixed position in the front of the sequence to retain the priority scheduling order. With a 30% probability, the crossover point is randomly selected to increase diversity.

[0034] An electronic device according to the present invention includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the computer program is loaded into the processor, it implements the method for joint scheduling of UAV release based on an improved genetic algorithm in any one of the above.

[0035] A storage medium according to the present invention stores a computer program. When the computer program is executed by a processor, it implements the method for joint scheduling of UAV release based on an improved genetic algorithm in any one of the above.

[0036] Beneficial effects: Compared with the prior art, the present invention has the following remarkable advantages: Combining the improved genetic algorithm and the joint scheduling mode: The present invention innovatively combines the improved genetic algorithm with the hangar deck joint scheduling mode, effectively solving the problems of the efficiency of UAV scheduling and resource utilization in complex environments. The improved genetic algorithm significantly improves the convergence speed of the algorithm and the quality of the scheduling solution by introducing strategies such as dynamic mutation rate control strategy and local search. Through the improvement of population initialization, crossover, mutation and selection operations, the balance between global search and local optimization of the algorithm is ensured, the trouble of local optimal solutions is reduced, and thus the overall efficiency and solution quality of UAV scheduling are greatly improved.

[0037] Dynamic collision detection and path optimization: For the hangar scheduling problem, the present invention adopts a path planning method based on a dynamic collision detection algorithm, ensuring the safety between UAVs during the dynamic process by calculating the UAV paths in real time and applying linear interpolation, avoiding the dynamic collision problems that may be ignored by traditional static path planning methods. By considering the threat areas and real-time positions of each UAV in path planning, the UAV scheduling scheme is optimized, reducing conflicts and waiting time in path planning.

[0038] Joint scheduling of hangar and deck: The joint scheduling mode proposed by the present invention integrates the hangar and deck scheduling tasks through a time synchronization mechanism, optimizing the connection of the two-stage tasks. Especially in terms of the time coupling between Task 1 (hangar scheduling) and Task 2 (deck scheduling), a joint scheduling method is adopted to ensure the seamless connection of the two-stage tasks, effectively avoiding resource conflicts and idle time, and thus significantly improving the resource utilization rate and overall scheduling efficiency.

[0039] Optimizing resource allocation and scheduling time: By introducing scheduling optimization for refueling trucks, ammunition carriers and tractors in deck scheduling, the order of refueling and ammunition loading tasks is optimized, reducing the waiting time caused by resource conflicts. Through an intelligent resource allocation strategy, it is ensured that each UAV can efficiently complete refueling, ammunition loading and towing tasks, thereby improving the takeoff efficiency of UAVs. Especially in emergency tasks, the number of takeoff sorties can be maximally increased.

[0040] Verification by simulation experiment: Through simulation verification, the joint scheduling mode of the present invention reduces the total scheduling time by about 16% compared with the independent scheduling mode, and under the same time requirements, the number of UAV sorties released increases significantly, with the takeoff sorties increasing by 33.3% to 60% and 60% to 300% respectively. Compared with the genetic algorithm and the particle swarm algorithm, the scheduling results of the improved genetic algorithm show higher efficiency. Especially in the case of short task time requirements, it can greatly increase the takeoff sorties and meet the high-efficiency scheduling requirements in actual scenarios.

[0041] Through these innovations, the present invention not only achieves significant improvements in the efficiency of UAV scheduling and resource utilization, but also greatly enhances the stability of the algorithm and the quality of the scheduling solution, and can meet the actual needs of carrier-based UAV scheduling and launching in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is a flowchart of the present invention;

[0043] Figure 2 is a schematic diagram of collision detection of the present invention;

[0044] Figure 3 is a diagram of the hangar station of the present invention;

[0045] Figure 4 is a schematic diagram of the aircraft carrier deck of the present invention;

[0046] Figure 5 is a coordinate station diagram of the hangar of the present invention;

[0047] Figure 6 is a deck station diagram of the present invention;

[0048] Figure 7 is a comparison of the results of eight operations of hangar scheduling of the present invention;

[0049] Figure 8 is a comparison of the optimal solution times of three algorithms for hangar scheduling of the present invention

[0050] Figure 9 is a comparison of the results of eight operations of deck scheduling of the present invention

[0051] Figure 10 Comparison of the optimal solution times of three deck algorithms

[0052] Figure 11 is the working path of the refueling vehicle of the present invention;

[0053] Figure 12 is the working path of the ammunition carrier of the present invention;

[0054] Figure 13 is the driving path of the combined scheduling tractor of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0055] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0056] As Figures 1-13 shown, an embodiment of the present invention provides a UAV launching combined scheduling method based on an improved genetic algorithm in a complex environment, including the following steps:

[0057] (1) In the hangar stage, a drone hangar outbound scheduling model is established. By using an improved collision detection method combined with linear interpolation, a real-time safety assessment is carried out for the whole process of the drone from the initial parking position to the deck parking position. If the collision detection finds that the following formula does not hold, the scheduling plan is not feasible:

[0058] |O1O2|>R1+R2+d

[0059] Where: O1 and O2 respectively represent the center coordinates of different drones, and R1 and R2 are the corresponding safety boundary radii of the two, then it is determined that the scheduling plan conflicts and is not feasible at this moment.

[0060] When using the drone outbound scheduling modeling, consider the following constraints: the initial positions and moving speeds of the drones, tractors, and elevators are constants; at the same time, any one tractor or elevator can only serve one drone; it is prohibited for multiple tractors to tow the same drone in parallel; the tractors process the drones they are responsible for in the set order and are not allowed to skip randomly in the task list; the processing capacity of any one tractor or elevator within a scheduling cycle is limited, denoted as and, if there is or does not meet the requirements, it is determined that the scheduling cannot continue; the parking positions cannot be occupied simultaneously;

[0061]

[0062] Among them, x ij represents that the i-th drone is processed by the j-th tractor, and y jk represents that the i-th drone is processed by the k-th elevator, and M t and M l are the maximum processing quantities of the tractor and the elevator respectively.

[0063] The whole process of the drone from the initial position out of the warehouse to the deck parking position aims to minimize the total duration, and it is necessary to comprehensively balance the scheduling order of the tractor and the elevator and the waiting time that the drone may generate.

[0064] The objective function is to minimize the total time for all drones to reach the final parking position from the initial position. Specifically, the objective function can be expressed as:

[0065]

[0066] Among them, T i represents the total time required for the i-th drone to reach the final parking position from the initial position, including the following parts: the time T t1 for the tractor to reach the drone position, the time T t2 for the hangar tractor to tow the drone to the elevator position, and the elevator ascending / descending time T l, the time T for the deck tractor to tow the UAV from the towing area position to the parking position on the deck t3 , overall, the total time T of the i-th UAV i can be expressed as:

[0067] T i = T t1 (j,i) + T t2 (i,k) + T l + T t3 (a,m)

[0068] where: j and k respectively represent the selected tractor and elevator indices, j = 1, 2. m represents the selected parking position index, m = 1, 2......8. a represents the selected towing area index, a = 1, 2. distance(a,b) represents the distance between position a and position b.

[0069] (2) Design an improved genetic algorithm to optimize population initialization, crossover, and mutation operations, introduce a dynamic mutation rate and a local search strategy to improve the algorithm's convergence efficiency and the quality of the scheduling solution, as follows:

[0070] Individual initialization: Adopt a hybrid strategy of "random generation + knowledge assistance" to balance the diversity and directionality of solutions in the early stage. The specific steps are as follows:

[0071] 1. Initialize the population size to N, thus generating N initial solutions;

[0072] 2. For each newly generated individual:

[0073] With a 70% probability, use a simple random permutation method to make the order of UAVs in the scheduling sequence completely random; with a 30% probability, construct the individual based on domain experience. Specifically: randomly select a starting position among the positions close to the tractor and elevator in order to obtain a better initial solution, and then randomly permute the positions of the remaining UAVs to complete the entire sequence.

[0074] Crossover operation: Introduce a multi-point crossover method, and distribute the crossover point selection probability according to the following rules: There is a 70% probability of concentrating the crossover points in the front part of the sequence to maintain the relatively good gene arrangement in the scheduling leading individuals, and the remaining 30% probability randomly distributes the crossover points at any position in the sequence to break the existing order and increase the diversity of gene recombination. In this way, certain excellent segments (such as the order of processing certain UAVs first) can be well retained, and the algorithm will not be stuck in a single mode, preventing it from falling into a local optimum.

[0075] Mutation operation: A linear dynamic mutation rate mechanism from low to high is introduced to ensure that the algorithm focuses more on local fine search in the initial stage and gradually improves the ability to jump out of local extreme values in the later stage. The initial mutation rate is set to 0.03, and the final mutation rate is set to 0.1. The dynamic mutation rate adjustment formula is:

[0076]

[0077] where P mutation represents the current mutation rate, P initial and P final are the initial and final mutation rates respectively, call_count is the number of calls to the mutation function, and max_calls is the assumed maximum number of calls. The dynamic mutation rate balances the global exploration and local exploitation capabilities through linear growth. Specifically, the initial mutation rate (such as 0.03) gives the algorithm a larger search space, which helps to refine the search; while the final mutation rate (such as 0.1) increases the possibility of jumping out of the local optimum. At the same time, the range of the mutation operation is restricted to avoid mutating the first unmanned aerial vehicle (usually the priority scheduling target), and only random exchanges are performed on other unmanned aerial vehicles. This method effectively balances the global exploration and local exploitation capabilities of the algorithm, ensures the efficiency and rationality of the scheduling order, and improves the overall optimization performance of the shipborne unmanned aerial vehicle scheduling.

[0078] Selection operation: An elitist retention mechanism is adopted to ensure that several individuals with the highest fitness in the previous generation directly enter the next generation; at the same time, the selection of the remaining individuals is divided into two categories: 50% roulette wheel and 50% tournament, in order to balance probability and competitiveness. To further highlight the importance of the unmanned aerial vehicles near the tractor and elevator, in this invention, when performing tournament selection, a higher winning weight is given to those individuals whose initial positions are closer to the tractor or elevator, making them more likely to enter the next generation population, thereby improving the overall scheduling efficiency and accelerating the algorithm convergence speed.

[0079] Fitness calculation: The quality of an individual is measured by calculating the total unmanned aerial vehicle scheduling time. The formula is as follows:

[0080]

[0081] (3) Establish a path scheduling model for deck functional vehicles with the goal of minimizing the UAV take-off time and the waiting time for resource conflicts. For the take-off scheduling problem of UAVs on the deck, a scheduling model is established with the goal of minimizing the total deck scheduling time of UAVs. By analyzing the processes of refueling, loading ammunition, and towing UAVs to the take-off position on the deck, considering the constraint conditions of various resources, an improved genetic algorithm is used for optimization to solve the optimal refueling and ammunition loading sequences, as well as the total scheduling time of the entire process. The assumptions of the scheduling model are as follows: The types and quantities of resources are limited. By default, only one refueling vehicle, one ammunition loading vehicle, and one tractor are configured. Refueling and ammunition loading are mutually exclusive processes. If a vehicle arrives and finds that the UAV is occupied by another functional vehicle, the later vehicle needs to wait. The planar movement speed of functional vehicles in the deck area is fixed at 30 (unit distance / minute). The refueling operation takes 3 minutes, and the ammunition loading operation takes 5 minutes. There are two available take-off positions on the deck, and the status of the take-off device is marked and managed by the array Fly_statue.

[0082] The constraint conditions of the scheduling model are as follows: Any UAV can only be towed after all pre-departure guarantees (refueling, ammunition loading) are completed, and refueling and ammunition loading are only performed once each.

[0083]

[0084] Where δ i,j is an indicator function that takes 1 when x i = j and 0 otherwise.

[0085] It is prohibited for the refueling vehicle and the ammunition loading vehicle to serve the same UAV at the same time. If the target UAV is occupied during vehicle movement, it needs to wait. Before any guarantee process is started, it is necessary to ensure that the relevant resources are available. Only one UAV can be accommodated at each parking position or take-off position at the same time. If the required resources are not ready, it must wait. Only UAVs that have completed refueling and ammunition loading can be towed to the take-off position. The two take-off devices on the deck cannot provide take-off services for multiple UAVs at the same time.

[0086] The objective function of the scheduling model is as follows:

[0087] The objective function is to minimize the waiting time for refueling and ammunition loading conflicts:

[0088]

[0089] Where: W oil,i is the waiting time of the refueling vehicle for the i-th UAV. W bomb,i is the waiting time of the ammunition loading vehicle for the i-th UAV. The scheduling model is established as follows: UAV set: A = {1, 2,..., M}, refueling task sequence: OilTasks = [o1, o2,..., o M, Bomb loading task sequence: BombTasks = [b1, b2,..., b M , UAV status: AirsStatus[i] ∈ {-1, 0, 1}, where: -1 indicates refueling, 0 indicates idle and can accept tasks, 1 indicates bomb loading. UAV towing demand status: AirsNeedDrags[i] ∈ {0, 0.5, 1.0}, where: 0 indicates that refueling and bomb loading are not completed, 0.5 indicates that refueling or bomb loading is completed, 1.0 indicates that refueling and bomb loading are completed and waiting for towing. Time calculation: The vehicle movement time is determined by the Euclidean distance, and the operation times are 3 minutes for refueling and 5 minutes for bomb loading respectively:

[0090]

[0091] Waiting time processing: If the UAV is still being served by other vehicles when the vehicle arrives, record the waiting until the operation is completed; Conflict handling: Introduce a scheduling function to simulate the queuing access process of refueling vehicles and bomb loading vehicles, and update the total waiting time and the overall scheduling time of the UAV caused by resource conflicts in real time.

[0092] Propose a combined scheduling mode for the hangar and the deck. Integrate the two-stage tasks through a time synchronization mechanism to optimize task connection and resource allocation; The core of the combined scheduling is to connect the hangar stage (Task 1) and the deck stage (Task 2) with a time synchronization mechanism: When the UAV completes towing in the hangar and arrives at the deck parking position, record its arrival time in real time; When the deck functional vehicle is ready to perform refueling or bomb loading tasks for the UAV, if it is detected that it has not arrived, the vehicle needs to wait; By sharing the connection time from the hangar to the deck, resource idle and vehicle conflicts can be reduced, and the overall cooperation efficiency can be improved.

[0093] (4) Verify the effect of the combined scheduling mode through simulation experiments, count the total scheduling time of the UAV, and analyze the optimization degree of the combined scheduling mode compared with independent scheduling. The experimental results show that the total scheduling time is reduced by 16%, and the takeoff sorties are increased by 33.3% to 60% and 60% to 300% respectively.

[0094] In order to verify that the improved genetic algorithm has better algorithm solution performance and faster convergence ability, and can better meet the scheduling planning time requirements and planning real-time requirements. Set up simulation experiments and take the results of running 8 times each.

[0095] (S1) Operating results of the hangar scheduling particle swarm algorithm

[0096]

[0097] (S2) Operating results of the hangar scheduling genetic algorithm

[0098]

[0099] (S3) Operation results of the improved genetic algorithm for hangar scheduling

[0100]

[0101] (S4) Operation results of the particle swarm algorithm for deck scheduling

[0102]

[0103]

[0104] (S5) Operation results of the genetic algorithm for deck scheduling

[0105]

[0106]

[0107] (S6) Operation results of the improved genetic algorithm for deck scheduling

[0108]

[0109] (S7) To achieve higher efficiency in UAV scheduling, joint scheduling of hangar and deck UAVs is adopted, and the operation results are as follows:

[0110]

[0111]

[0112] (S8) The number of take-off sorties in the same time The following table compares the total number of sorties of UAVs released by different algorithms under different time requirements.

[0113]

[0114] Through Figure 7From the operation results, it can be seen that the optimal feasible solution of the improved genetic algorithm in hangar scheduling simulation is 30.82, the minimum number of iterations is 2 times, and the average number of iterations is 3 times, which is significantly better than 32.04 of the genetic algorithm, the minimum number of iterations is 4 times, and the average number of iterations is 6.25 times, with increases of 3.96%, 100% and 108.3% respectively; it is also significantly better than 32.91 of the particle swarm algorithm, the minimum number of iterations is 9 times, and the average number of iterations is 4.87 times, with increases of 6.78%, 200% and 62.3% respectively. The comparison results show that the improved genetic algorithm improves the quality of population initialization through knowledge-based population initialization, and the crossover strategy and optimized mutation strategy that combine randomness and determinacy make it have better feasible solutions while achieving fewer and more stable optimal iteration times, proving that the improved genetic algorithm has better algorithm solution performance and faster convergence ability, and can better meet the planning time requirements and algorithm real-time requirements of scheduling.

[0115] Through the comparison of experimental results, the efficiency differences between independent scheduling and joint scheduling methods are analyzed, and the advantages of joint scheduling in optimizing scheduling time and reducing resource conflicts are proved.

[0116] Under the independent scheduling mode based on the improved genetic algorithm, hangar scheduling and deck scheduling are carried out independently, and the total scheduling time is the sum of the time consumed by both. From the operation results, it can be seen that the best total time for unmanned aerial vehicles (UAVs) to be scheduled from the hangar to the deck is 30.82, the best total scheduling cost time for deck UAVs is 50.38, and the total time spent is 81.2. In contrast, through synchronous planning of the two-stage tasks, joint scheduling enables seamless connection of tasks and effectively reduces the scheduling time of the entire task. Experimental results show that the best total scheduling cost time for hangar-deck UAV joint scheduling is 68.09, and the total joint scheduling time is reduced by about 16% compared with independent scheduling.

[0117] In the independent scheduling mode, due to the lack of coordination between tasks, problems such as idle resources (such as tractors and functional vehicles) may occur. By comprehensively considering the time coupling of Task 1 (hangar scheduling) and Task 2 (deck scheduling), joint scheduling realizes the dynamic allocation and reasonable utilization of resources, and significantly improves resource utilization.

[0118] The number of UAV sorties released by the improved genetic algorithm under the same time limit is significantly higher than that of the traditional genetic algorithm and the particle swarm algorithm. When the specified times are 55, 60, 65, and 70, the number of sorties released by the improved genetic algorithm increases by 100%, 66.7%, 40%, and 33.3% compared with the genetic algorithm, and increases by 300%, 66.7%, 75%, and 60% compared with the particle swarm algorithm. Within the same specified time, the improved genetic algorithm can release more UAVs. Especially in the case of more urgent task times, as the release time decreases, the growth rate of the release sorties generally shows an obvious upward trend.

[0119] This result shows that the joint scheduling of the improved genetic algorithm can not only improve the release efficiency of UAVs, but also, in emergency situations, maximize the number of released flights and enhance the task execution ability by reasonably optimizing the scheduling order and resource allocation.

Claims

1. A UAV release joint scheduling method based on an improved genetic algorithm in a complex environment, characterized in that It includes the following steps: (1) In the hangar stage, construct a UAV hangar scheduling model, and adopt a method combining dynamic collision detection and linear interpolation to detect spatial conflicts in real time during the process of UAVs leaving the hangar; dynamic collision detection updates the UAV positions moment by moment to determine whether the threat circles of any two UAVs overlap. When the following conditions are met, it is determined as a conflict: |O1O2|>R1+R2+d Where: O1 and O2 respectively represent the center coordinates of different UAVs; R1 and R2 are the corresponding safety boundary radii of the two, and d is the redundant safety distance; (2) Based on the improved genetic algorithm, globally optimize the UAV scheduling order. The improved genetic algorithm includes the following steps: (2a) Population initialization: Adopt a mixed strategy of random generation and knowledge assistance. 70% of the individuals are randomly arranged, and 30% of the individuals generate the initial sequence based on the principle of proximity to the tractor and elevator; (2b) Dynamic mutation rate control: The mutation rate increases linearly with the number of iterations. The formula is: Among them, P mutation represents the current mutation rate, P initial and P final are the initial and final mutation rates respectively, call_count is the number of calls to the mutation function, and max_calls is the assumed maximum number of calls; (2c) Selection operation: Adopt a combined strategy of elitist retention and tournament, and assign higher selection weights to the individuals close to the tractor and elevator; (3) In the deck stage, establish a functional vehicle path scheduling model with the goal of minimizing the waiting time of the refueling vehicle and the ammunition loading vehicle, and satisfy the following constraints: refueling and ammunition loading are mutually exclusive processes, and the later vehicle needs to wait; the moving speed of the functional vehicle is fixed at 30 unit distances per minute, the refueling time is 3 minutes, and the ammunition loading time is 5 minutes; the objective function is: Where: W oil,i is the waiting time of the refueling vehicle for the i-th UAV. W bomb,i is the waiting time of the ammunition carrier vehicle for the i-th UAV; (4) Connect the hangar and deck stages through a time synchronization mechanism, and record in real time the time when the UAV arrives at the deck parking position. If the deck functional vehicle detects that the UAV has not arrived, it will enter the waiting state to achieve coordinated scheduling of resources in the two stages.

2. The method for joint scheduling of UAV release based on improved genetic algorithm in complex environment according to claim 1, wherein The hangar scheduling model needs to satisfy the following constraints: (a) At the same moment, the tractor or elevator only serves one UAV; (b) The tractor processes tasks in a fixed order, and skipping or parallel operations are prohibited; (c) The handling capacity of the tractor and the lift is limited by the maximum handling quantity M and M l , satisfying: where x ij indicates that the i-th drone is handled by the j-th tractor, and y jk indicates that the i-th drone is handled by the k-th elevator, and M t and M l are the maximum handling quantities of the tractor and the elevator respectively.

3. The method for joint scheduling of UAV release based on improved genetic algorithm in a complex environment according to claim 1, wherein, The fitness function of the improved genetic algorithm is: Among them, the total scheduling time is the sum of the completion times of all UAVs from the hangar to the deck parking position.

4. The method for unmanned aerial vehicle release joint scheduling based on improved genetic algorithm in a complex environment according to claim 1, wherein In the deck scheduling model, the UAV status is defined by the following variables: AirsStatus[i]∈{-1,0,1}, where: -1 means refueling, 0 means idle and can execute new tasks, 1 means loading ammunition; AirsNeedDrags[i]∈{0,0.5,1.0}, where: 0 means the guarantee is not completed, 0.5 means the guarantee is completed, 1.0 means all guarantees are completed.

5. The method for joint scheduling of UAV release based on improved genetic algorithm in a complex environment according to claim 1, characterized in that The time synchronization mechanism specifically includes: when the UAV completes the hangar towing and arrives at the deck parking position, update its arrival time T arrival ; the deck functional vehicle starts to perform refueling or ammunition loading tasks after T arrival ; if the functional vehicle arrives early, it will wait until T arrival .

6. The method for joint scheduling of UAV release based on improved genetic algorithm in a complex environment according to claim 1, wherein, In dynamic collision detection, the UAV path planning adopts the linear interpolation method, updates the position every 0.1 seconds, and calculates the overlapping state of the threat circles.

7. The method for joint scheduling of UAV release based on improved genetic algorithm in a complex environment according to claim 1, characterized in that, The deck scheduling model also includes takeoff position management. The takeoff position status is dynamically marked by the array ˋFly_statusˋ, and only one UAV is allowed to use the takeoff position at the same moment.

8. The method for joint scheduling of UAV release based on improved genetic algorithm in a complex environment according to claim 1, wherein, The crossover operation of the improved genetic algorithm adopts the multi-point crossover strategy. With a 70% probability, it crosses at a fixed position in the front of the sequence, and retains the priority scheduling order. With a 30% probability, it randomly selects the crossover point to increase diversity.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the computer program is loaded into the processor, it implements a joint scheduling method for unmanned aerial vehicle (UAV) release based on an improved genetic algorithm in a complex environment according to any one of claims 1-8.

10. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements a joint scheduling method for unmanned aerial vehicle (UAV) release based on an improved genetic algorithm in a complex environment according to any one of claims 1-8.

Images