基于混合优化算法的多UUV协同侦察任务分配方法研究

Research on Task Assignment Method of Multi-UUV Cooperative Reconnaissance Based on Hybrid Optimization Algorithm

针对多UUV系统静态任务规划过程中,任务分配与航路规划相对独立造成的总体方案次优性问题,将非支配排序遗传算法(Non-dominated Sorting Genetic AlgorithmⅡ,NSGA-Ⅱ)与动态规划算法相结合,提出一种混合优化算法。将NSGA-Ⅱ作为总体优化框架,为各UUV分配任务子集;将动态规划算法用于各UUV的任务子集,基于最短路径准则优化得到各UUV的任务序列;根据任务序列进行任务分配方案的评估与优选。基于典型想定进行仿真实验,结果表明,通过将动态规划算法嵌入NSGA-Ⅱ优化框架,在任务分配过程中显式地考虑任务执行顺序对方案性能的影响,能够提升寻优方案的质量,加快NSGA-Ⅱ寻优过程的收敛。

In order to alleviate the sub-optimal problem of the scheme,caused by the relative independence of task assignment and route planning,in the process of static task planning of multi-UUV,a hybrid optimization algorithm is proposed by the means of combining non dominated sorting genetic algorithmⅡ(NSGA-Ⅱ)with dynamic programming algorithm.NSGA-Ⅱis used as the overall optimization framework to assign tasks to each UUV.Then,the dynamic programming algorithm is applied to each UUV task subset to obtain the task sequence based on the shortest path criterion,which is used evaluate and optimize the task allocation scheme.The simulation results based on typical scenarios show that by embedding dynamic programming algorithm into NSGA-Ⅱoptimization framework and explicitly considering the influence of task execution order on the performance of the scheme in the process of task assignment,the quality of the optimization scheme can be improved and the convergence of NSGA-Ⅱoptimization process can be accelerated.