摘要
进化规划算法中变异是唯一的操作,因此变异算子对进化规划算法的性能有决定性的影响。文中以高斯变异算子为例,研究了变异算子在进化进程的作用,分析了进化规划算法不收敛的原因以及变异算子与进化代数、收敛精度间的关系。对传统进化规划算法和多群进化规划算法的性能进行了仿真研究,仿真结果表明了分析结果的正确性。
Mutation is the exclusive operation in evolutionary programming （EP） algorithm, thus the performance of EP algorithm is definitively determined by mutation operator. Taking the example of Gauss normal mutation operator, the effect of mutation operator in evolution process is presented, and the reasons of divergence of EP algorithm, relationships between mutation operator and evolutionary generations, relationships between mutation operator and convergence precision are analyzed. Performances of the traditional evolutionary programming algorithm and multi-subgroup evolutionary programming （MEP） algorithm are simulated, and simulating result confirms the validity of analysis.
出处
《海军工程大学学报》
CAS
北大核心
2007年第1期48-52,共5页
Journal of Naval University of Engineering
关键词
进化规划
收敛
性能分析
evolutionary programming
convergence
performance analysis
