基于收缩精度的遗传算法逐级进化策略

为了有效克服传统遗传算法主观设定进化代数的弊端并提高算法进化期间的搜索效率,根据控制论中的反馈控制机理,通过适应度函数值的分散程度定义了收缩精度,并按照收缩精度将算法的进化期划分为不同的3个时期。在不同的进化期,采用不同形式的适应度函数以加大种群内个体之间的差异度。对交叉算子进行了改进,采用相关性配对交叉与改进的自适应交叉概率相结合的交叉算子,使算法达到较快的收敛速度。最后的算例表明,改进的遗传算法科学有效。

Multi-step evolution method in genetic algorithms based on contract precision

T o ef fect ively overcome the disadvantages of t radit ional g enetic alg orithm w hose terminate r ule w as desig ned subject ively and enhance the search ef ficiency dur ing the evo lut ion period, the definitio n of co nt ract precision w as presented acco rding to the feedback cont rol principle, and then the evo lut ion perio d w as divided into three part s based o n cont ract precision. T o enlarge the difference in individuals, different fitness metho ds were ado pted in different phases. And the cross o perator w as impro ved w ith auto-adapt ive cr ossover probability and uncorr elated matching to r ealize a hig her convergence speed. The calculat ion result s show that the impr oved GA is more scientif ic and eff icient