摘要
针对传统差分进化算法在求解作战目标分配时存在的参数敏感性高、高维变量求解精度低和易陷入局部最优等缺陷,提出一种改进的多策略协同差分进化算法。以精英种群引导3个等规模种群协同进化,各种群将产生优秀变异体的历史信息作为自学习经验,根据进化程度适应性选择变异策略、缩放因子和交叉概率,从而平衡全局搜索能力和收敛速度。与3种常用算法在8个测试函数上进行实验对比,对最优解的平均值、标准差、Wilcoxon rank-sum检验和胜率进行分析,测试了算法的收敛性和稳定性。以某次联合火力打击为例进行仿真实验,结果表明,算法能够有效解决作战目标分配问题。
Aiming at the shortcomings of traditional differential evolution algorithm in solving operational targets assignment,such as high parameter sensitivity,low accuracy of high-dimensional variables and easy to fall into local optimization,an improved multi strategy cooperative differential evolution algorithm is proposed.The elite population is used to guide the three coevolution populations which have the same size.Each population takes the historical information of excellent variants as self-learning experience,and adaptively selects mutation strategy,mutation scaling factor and crossover probability according to the degree of evolution,so as to balance the global search ability and convergence speed.Compared with three common algorithms on eight test functions,the average value,standard deviation,Wilcoxon rank-sum test and winning rate of the optimal solution are analyzed to test the convergence and stability of the algorithm.Taking a joint fire strike as an example,the simulation results show that the algorithm can effectively solve the problem of operational targets assignment.
作者
马悦
吴琳
郭圣明
MA Yue;WU Lin;GUO Sheng-ming(National Defense University,Beijing 100091;Unit 31002 of PLA,Beijing 100091,China)
出处
《指挥控制与仿真》
2022年第4期31-41,共11页
Command Control & Simulation
关键词
作战目标分配
差分进化
多策略
协同进化
自适应参数
operational targets assignment
differential evolution
multi strategy
coevolution
adaptive parameter