摘要
典型的进化策略受自然进化过程的启发而成为求解全局优化问题的重要方法 .传统的 ES变异算子作为一个主要的进化技术是建立在正态分布的随机变量基础上的 .本文提出了基于指数分布的进化策略由于采用了新的变异算子有效地减少了产生探试解的成本 。
The canonical evolutionary strategy is a global optimization problem solver, which gets inspirations from natural evolutionary processes. As a major technique of evolution, the mutation operator of the classical ES is based on the normal distribution random variable. It is shown in this paper that exponential distributin based evolutionary strategy is superior to the classic one in most cases due to that the new operator can efficiently reduce the cost of the trial solution generation compared with the traditional one.
出处
《纯粹数学与应用数学》
CSCD
2003年第1期16-21,共6页
Pure and Applied Mathematics
关键词
全局优化
进化策略
变异算子
指数分布
正态分布
global optimization, evolutionary strategies, mutation operator, exponential distribution, normal distribution
