The Study Of Multiple UAVs Routing Planning For Delivery Based On Evolutionary Multi-objective Optimization

With the rapid development of the new generation of information technology,especially artificial intelligence technology,UAVs are widely used in various fields.In the military field,UAVs can conduct real-time reconnaissance and fire strikes on targets while effectively avoiding casualties.At the same time,they can break through the constraints of complex terrain environments and achieve rapid supply of weapons,ammunition,and medical drugs.In the civilian field,the delivery of packages by UAVs in the last mile can effectively reduce the logistics cost of enterprises.In particular,contactless distribution based on UAVs is also crucial in the post-epidemic era.In view of this,it is of great significance and value to carry out multi-UAVs delivery path planning.Focusing on the path planning of multi-UAVs distribution,the paper firstly analyzes the actual needs and application background of multi-UAVs performing battlefield material support tasks,and summarizes the current research status of UAV path planning models and optimization algorithms at home and abroad.On this basis,starting from the two goals of the total flight distance of UAVs and the total number of UAVs,carry out research on multi-UAVs distribution and multi-target path planning for battlefield material support.The main research work of the paper is as follows:Firstly,for multi-UAVs distribution path planning under the large-scale mission nodes,the total flight distance and total number of UAVs are minimized under the constraints such as the hard time window and maximum load limit of UAV to support mission nodes.In order to solve the model quickly and accurately,a parallel decomposition multi-objective optimization algorithm p MOEA/D-DP is designed based on the co-evolution of double-population and double-crossover operators.The dualpopulation co-evolution mechanism utilizes the useful information of infeasible solutions in the auxiliary population,which effectively improves the convergence speed of the algorithm.What’s more,under the guidance of the idea of parallel and co-evolutionary efficiently search for the objective space partition,the parallel design of the algorithm is completed based on the MPI programming interface.The experimental results show that p MOEA/D-DP can effectively solve the optimization problem of large-scale task nodes while significantly reducing the computational time cost.Secondly,for the multi-UAVs distribution path planning under the dynamic change of mission nodes,the concept of critical time points is introduced,and so that the dynamic problem is transformed into a series of static optimization problems.Next we formulated a two-stage optimization strategy of "initial path pre-optimization plus critical time point path re-optimization",and designed a dual-population co-evolution constrained optimization algorithm(CMOEA-c)and an adaptive large-field search algorithm(ALNS)composed of Two-stage hybrid intelligent optimization algorithm.The experimental results show that CMOEA-c can search for a high-quality initial optimization path in the initial stage.In the dynamic re-optimization stage,using ALNS can provide a high-quality path solution while meeting the high timeliness requirements of the battlefield environment.Its performance is better than the existing classical algorithms.