摘要
提出一种新的求解函数优化的快速演化算法;新算法的特征是引入一种基于高斯变异和Cauchy变异的混合自适应变异算子,并作为算法的唯一遗传算子;提出多父体变异的群体爬山搜索策略;采用随机排序选择策略,克服了经典算法易于陷入局部最优解的常见弊病;新算法具有保持群体的多样性、全概率收敛、淘汰压力小、子空间搜索、快速收敛、评价次数少等特性;通过7个标准测试函数测试结果表明,新算法在所有的测试函数中体现出很好的性能,具有稳定、高效和快速等特点。
A novel and fast evolutionary algorithm （NFEA） for function optimization problems is proposed. It has some new features, such as introducing a hybrid mutation operator based on Gaussian mutation and Cauchy mutation, using multi- parent mutation＇s colony mountain climbing search strategy and stochastic ranking strategy, which conquered the shortcomings where classic algorithms are easy to fall into the local optimum. The characteristics of the proposed algorithm include to keep the variety of colony, one probability convergence. And the selection pressure is small. Moreover, the number of function evaluations are also less than other compared algorithms. The new algorithm is tested on 7 benchmark functions, the resuits indicate that the new algorithm is stable, effective and fast; and it can solve all of the tested functions very well.
出处
《计算机测量与控制》
CSCD
2007年第5期654-656,659,共4页
Computer Measurement &Control
基金
"十一五"民用航天项目(C5220061318)
湖北省人文基地项目(2004B0011)
湖北省自然科学基金资助项目(2003ABA043)
关键词
演化算法
函数优化
混合变异
随机排序
evolutionary algorithm
function optimization
hybrid mutation
stochastic ranking
