Research On Cooperative Target Allocation Algorithm Of Multi-UAV

Collaborative target allocation of multiple UAVs is a hot trend in the development of UAVs technology.The assignment between multiple UAVs and targets has great influence on the flexibility and efficiency of the mission execution process.Depending on the level of synergies available,a single UAV would also be popular in actual UAV operations.Of course,the real battlefield environment is dynamic and changeable,and a single UAV performing various tasks will face multiple challenges.Target allocation is an important part of UAV planning and the key of UAV cooperative operation,which is of great research significance.Task assignment mainly consists of two parts,problem modeling and algorithmic solution of modeling content.This thesis mainly focuses on the modeling and solving of two aspects of objective assignment and analyzes the optimization model based on single objective and multi-objective.The main research contents and work include:The problem of single objective optimization model is studied.For single-target optimization,the TSP model,assignment model and transport model are respectively established according to whether the number of UAV and target is evenly distributed,considering only the distance cost,time cost or threat cost of UAV during reconnaissance or attack.Genetic simulated annealing algorithm and heuristic Hungarian algorithm are used to solve the three task models.In view of the disadvantage that simulated annealing algorithm is not good at global search when optimizing models,genetic evolution operation is carried out before individual solution operation.Aiming at the shortcomings of Hungarian algorithm in solving high-dimensional problems such as slow convergence speed and poor accuracy,the operation of population in heuristic algorithm is added in this thesis to improve the optimization effect of Hungarian algorithm.For the multi-force problem in the transportation problem,the number of UAV and target is not equal.It is impossible to deal with the non-square matrix problem by using the ordinary Hungarian algorithm.In this thesis,the Hungarian algorithm with edge complement is adopted.The multi-objective optimization model is studied.Aiming at the construction of multi-objective optimization model,the construction of objective function in most models does not comprehensively consider the loss cost part and the target return part in the actual battlefield.In this thesis,the cost of four parts is comprehensively considered.The target distribution model of multi-UAV is established based on the minimum flight range model,the minimum flight time model,the minimum loss cost model and the maximum target revenue model.It is easy to get into the local optimal solution to solve the problem of target assignment by genetic algorithms.In this thesis,the hybrid optimization genetic algorithm and the greedy algorithm are put forward to improve the local search ability of typical genetic algorithms.Meanwhile,the selection operator is improved to use the disaster algorithm on individuals,the crossover operator mapping rule is improved,and the gene is used to enrich the diversity of individuals and avoid getting into the local optimal solution.