According to benchmark complex optimization problem, and puts forward the center based on adaptive mutation and crossover probability of differential evolution algorithm--center differential evolution (variation center mutation-based differential, CMDE) algorithm is proposed to explain. The algorithm firstly improvement of the individual variation form the current generation, namely the group as the center, according to the variation vector in three random individual vector of the fitness function between the size relations, to determine the direction of the poor vector; Then give the adaptive crossover probability strategies that cross function through based on analysis of the function between individual vector to value in the distribution of internal group, to make sure that each individual crossover probability. Through several Benchmark function test showed that, CMDE algorithm has higher convergence speed, and to Benchmark complex problems of high precision, optimal performance.
針對(duì)高維復(fù)雜優(yōu)化問題,提出了基于中心變異和自適應(yīng)交叉概率的差分進(jìn)化算法———中心變異差分進(jìn)化(center mutation-based differential evolution, CMDE)算法。該算法首先改進(jìn)了個(gè)體的變異形式,即把當(dāng)前代的群體中心作為基向量,依據(jù)參加變異的三個(gè)隨機(jī)個(gè)體向量間的函數(shù)適應(yīng)值的大小關(guān)系,確定差向量的方向;然后給出了自適應(yīng)交叉概率策略,即依據(jù)交叉的作用,通過分析個(gè)體向量間的函數(shù)適應(yīng)值在群體內(nèi)部的分布情況,確定每個(gè)個(gè)體的交叉概率。通過幾個(gè)Benchmark函數(shù)的測(cè)試表明,CMDE算法具有較快的收斂速度,且對(duì)于高維復(fù)雜問題的求解精度高,尋優(yōu)性能好。