Multi Objective Optimization Algorithm For UAV Group Task Allocation

With the rapid development of unmanned system,people have higher and higher requirements for the effectiveness and diversity of UAV group task allocation scheme.However,under the influence of COVID-19,the problem of single scheme in UAV group task allocation is more prominent.Once the obtained scheme can not be carried out,it needs to be customized again,resulting in a lot of time waste.Therefore,a multi-objective comprehensive learning set-based particle swarm optimization algorithm based on set decomposition is proposed uses the MOEA / D framework,the algorithm combines set based particle swarm optimization(S-PSO)and comprehensive learning particle swarm optimization(CLPSO).Without setting the super parameters of each target in advance,it can quickly and adaptively obtain more diverse feasible task allocation schemes according to multiple objectives,aiming at solving a group of feasible task allocation schemes as standby Program.The experimental results show that the algorithm can obtain a set of feasible task allocation schemes,and the algorithm has good convergence,diversity and stability,and has good practical significance in practical application.This paper mainly includes the following three parts1)In this paper,the task allocation problem of UAV group is reduced to vehicle routing problem with time window(VRPTW).At the same time,according to the characteristics of user time window in VRPTW,a pair of conflicting objective functions are proposed innovatively to balance the waiting time between UAV and user,so as to make the task allocation scheme more in line with the actual situation.2)This paper uses a two-stage algorithm to solve the Vehicle routing problem with time windows problem(VRPTW).In the first stage,the random nearest neighbor hybrid algorithm is used as the first stage to obtain the initial feasible task allocation scheme as the initial particles in the second stage.In the second stage,the MOEA/D framework is innovatively introduced into CS-PSO algorithm,and the improved MOCS-PSO/D framework is proposed to balance the weight of multiple objective functions in multi-objective optimization rapidly and adaptively,and the speed update formula is improved.At the same time,the local search strategy is improved by using graph based method,adding external storage set update strategy and adding adaptive stop policy It is omitted.The experimental results show that the algorithm has fast convergence speed and strong stability,and can obtain a variety of feasible task allocation solutions.3)Through the data analysis of the experimental results of two-stage MOCSPSO/D algorithm,the optimization objective based on preference is added,and a twostage MOCS-PSO/D algorithm based on preference is proposed,which makes the feasible task allocation scheme obtained by the algorithm more practical.