Abstract: Traditional evolutionary algorithm sees each chromosome of the population as independent individuals, and doesn’t make fully minning between its internal relations. In order to solve this problem, this paper proposes an evolution algorithm based on chromosome difference and variation attenuation principle. It uses the chromosome index and chromosome structure to do cross operation, gives experience crossover probability formula, set blind date frequency for each chromosome, according to the evolution law, variation attenuation principle is introduced into the mutation operator of evolutionary algorithm. Experimental results show that under 3 functions, the highest average evolution of this algorithm is 78, minimum search success rate is 99%, and is better than basic Genetic Algorithm(GA) and GA introduced division operator.

Key words: individual difference, Genetic Algorithm(GA), evolution, population, crossover, variation attenuation

摘要： 传统进化算法将种群中的每个染色体看作独立的个体，没有充分挖掘其内部关系。为解决该问题，提出一种基于染色体差异与变异衰减原则的进化算法。利用染色体的指标值和染色体的结构进行交叉运算，给出经验交叉概率公式，对每个染色体设置相亲次数，按照生物进化规律，将变异衰减原则引入到进化算法的变异算子中。实验结果表明，该算法 3种函数的平均进化代数最高为78，搜索成功率最低为99%，优于基本遗传算法和引入分裂算子的遗传算法。

关键词: 个体差异, 遗传算法, 进化, 种群, 交叉, 变异衰减

CLC Number: 

• TP18