随机选择变异及自适应差分进化算法的研究

Research on Random Selection Mutation and Adaptive Differential Evolution Algorithm

为兼顾搜索速度和精度,提高搜索效率,克服不易跳出局部最优的缺点,提出了随机选择变异策略、自适应调整变异率和自适应调整交叉率3种改进设想.利用3种改进设想改进DE算法,得到了7种改进算法,将7种改进算法应用到双容水箱液位模型闭环辨识案例中.结果表明,7种改进算法都提高了跳出局部最优的能力,综合比较可知,3种改进设想同时应用的改进算法性能最优.

In order to take into account search speed and search accuracy,improve search efficiency,overcome the shortcoming which does not easily escape from local optimum value,three improved assumptions are proposed,namely,random selection mutation strategy,adaptive adjustment of mutation rate and adaptive adjustment of crossover rate.Three kinds of improved assumptions are used to improve the basic algorithm and get seven kinds of improved algorithms,which are applied to two-tank water level model closed-loop identification case.Case verification shows that seven kinds of improved algorithms increase the ability to jump out of local optimum value.Comprehensive comparison shows that the performance of improved algorithm where three improved assumptions are applied simultaneously is the best.