基于改进遗传算法的多机协同多目标分配方法  

Multiple targets assignment of multiple UAVs' cooperation based on improved genetic algorithm

作　　者：王庆贺 万刚[2] 柴峥 李登峰 Wang Qinghe;Wan Gang;Chai Zheng;Li Dengfeng(66132 Forces,Beijing 100043,China;Information Engineering University,Zhengzhou 450001,China;61206 Forces,Beijing 100041,China)

机构地区：[1]66132部队 [2]信息工程大学 [3]61206部队

出　　处：《计算机应用研究》2018年第9期2597-2601,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(41301428;41371384)

摘　　要：针对复杂三维环境中多无人机协同多目标分配问题,在飞行代价函数建模的基础上,提出了一种改进遗传算法。首先通过引入启发式信息和采用随机生成的方法构造初始种群,保证了初始种群的多样性和高适应性;然后构造适应度函数,加入惩罚项排除不满足约束条件的方案;最后进行遗传操作,将变异产生的个体组成新的种群,把新种群中性能优异的个体加入到初始种群中,使初始种群个体种类更加丰富,扩大了解的范围。设计实验将改进遗传算法与基本遗传算法和差分进化算法进行了对比,实验结果表明,改进遗传算法在无人机与目标不同的数量关系下都能够得到合理的分配方案;改进遗传算法有效改善了早熟问题,并具有更快的收敛速度,适合于求解多无人机多目标分配问题。Aiming at multiple targets assignment of multiple UAVs’cooperation in the complex three dimensional environment,this paper proposed an improved genetic algorithm based on the modeling of flight cost function.Firstly,it generated the initial population by the introduce of heuristic information and the method of random construction,which ensured the diversity and high adaptability of the initial population;following it was to build the fitness function by adding a penalty to exclude schemes which did not satisfy the constraints;then came the genetic operations,where the mutate individual components to the new population,and it added the individual with excellent performance in the new population to the initial population,which made the initial population more abundant,and enlarged the scope of individuals.Experiment compared the proposed algorithm with basic genetic algorithm and differential evolution algorithm,and the experimental results show that the proposed algorithm is able to get a reasonable assignment on the condition of different UAVs and targets;the algorithm can effectively improve the problem of premature convergence,and it has a faster convergence speed,so it is suitable for solving multiple targets assignment of multiple UAVs.

关 键 词：多目标分配 多无人机协同 改进遗传算法 三维环境 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]