Research On The Multi-UAV Cooperative Task Assignment Method Based On The Swarm Intelligence Algorithm

The drone swarm combat is the main form of the future air warfare,including the execution of reconnaissance,attack,defense,etc,greatly reducing the cost of combat and improving the efficiency.The task assignment is the top-level planning of drone swarm combat and the core function of UAV command and control system.The essential purpose of the multi-UAV task assignment problem is to mainly consider the number of UAVs,battlefield environment,target value,single aircraft capability constraints and other factors,and to assign different missions to UAVs through optimization algorithms with the goal of optimal or sub-optimal overall mission efficiency.However,there are many issues yet to be improved,including the perfection of the mathematical model of UAV swarm task assignment,the global search capability of the mission assignment algorithm,etc.The thesis addresses a series of theoretical researches on the task assignment problem in the scene of multi-UAV attacking moving targets cooperatively with static obstacles.Specific results are as follows.1.To improve the optimization performance of the task assignment algorithm in the multi-UAV cooperative attack task assignment problem,a particle swarm task assignment algorithm SA-LRS-PSO based on simulated annealing(SA)mechanism and local random search(LRS)is proposed by combining the CMTAP model and particle swarm optimization(PSO).Firstly,a mathematical model is established in the consideration of the UAV range,maximum task completion time,task time window and task reward.According to the mission scenario,a single-chain particle coding and decoding algorithm is designed including the information of the UAV,target type and mission order.Otherwise,an adaptive learning factor and a social cognitive component are introduced to the particle updating process,while a local search initiation mechanism is designed based on SA,and mutation and crossover operators are used to perform the local random search to improve the global search capability of the algorithm.The result shows that for the multi-UAV cooperative attacking problem,the SA-LRS-PSO algorithm proposed in this paper can search for the global optimum without significantly increasing the computational consumption.2.In the mutli-UAV task assignment problem with obstacles,the current research cannot estimate the UAV range well,which leads to a certain deviation between the fitness function and the actual result in the task assignment algorithm.The factors mentioned above may cause a poor search ability of the task assignment algorithm.To address this problem,an evaluation index of UAV obstacle avoidance algorithm is designed for screening the UAV obstacle avoidance algorithm embedded in the multi-UAV cooperative task assignment algorithm.Firstly,a model is established about the path obstacles and obstacle avoidance problem.Then,an evaluation index of UAV obstacle avoidance algorithm is proposed considering the path length,computation time consumption and stability.The results show that the PSO algorithm has the superior performance,which is selected to be embedded in the multi-UAV cooperative task assignment algorithm.3.An improved PSO algorithm,TSL-PSO,is proposed for the multi-UAV cooperatively attacking moving targets task scenario with static obstacles.Firstly,the corresponding mathematical model and solution formula are established in the scenario of UAV chasing a moving target.Secondly,a meeting point solving method of UAVs and moving targets is proposed based on PSO and the bisection method for the UAV intercepting moving targets scenario with obstacles.In addition,the TSL-PSO algorithm is proposed based on the taboo-list(TL)-variable neighborhood descent(VND)algorithm and SA-LRS-PSO algorithm to generate a task scheduling scheme for multi-UAVs in parallel and participate in the particles’ iteration process to improve the population diversity and increase the convergence speed of the algorithm.The result shows that the TSL-PSO algorithm can realize the task assignment in this scenario and converge faster.The research above can improve the combat efficiency of the drone swarm and lay the theoretical and technical foundation for the intelligent application of the drone swarm in the modern battlefield.