摘要
为了改善遗传算法的收敛性能,提出了一种基于个体适应度的种群多样性度量函数,恰当地反映了遗传算法的进化阶段,预报了早熟收敛的趋势.设计了基于种群多样度函数的迁移算子和交叉算子,并对交叉、变异概率等进行了动态调整,构成了具有多层迁移特点的实数编码并行遗传算法.通过和其他优秀遗传算法对测试函数的验证比较,结果表明,该算法对于解决遗传算法中早熟、收敛速度慢等问题具有优越的性能.
In order to improve the convergence performance of genetic algorithms, a function measuring population diversity on the basis of the degree of individual adaptability was needed. This measuring function must reflect the evolutionary stage of the genetic algorithm and forecast premature trends appropriately. We designed a migration operator and crossover operator based on diversity of population functions, and made dynamic adjustments on crossover and mutation probabilities. This structured a parallel genetic algorithm with real coding and a multi-layer migration operator. Comparative experiments were made on benchmark functions. The results showed that this algorithm is obviously superior to other genetic algorithms in overcoming problems such as prematurity and slow convergence.
出处
《智能系统学报》
2008年第5期423-428,共6页
CAAI Transactions on Intelligent Systems
基金
黑龙江省自然科学基金资助项目(A200419)
关键词
遗传算法
种群多样性
迁移算子
实数编码
genetic algorithm
diversity of population
migration operator
real coding
