基于克隆进化选择算法的多目标优化研究

MULTI-objective optimization based on clonal evaluation selection algorithm

把免疫系统的克隆选择学说与生物进化法则应用到多目标优化计算中,引入免疫克隆学说的记忆单元体,使用聚类方法对其中的抗体进行不断的优化更新和劣体淘汰;采用非均匀变异操作促进种群抗体的多样性;通过抗体间亲和度体现种群中个体的竞争,抗体与抗原亲和度来抑制过度的竞争,维持种群广泛性。最后由计算机仿真实验,并与NSGA-Ⅱ算法比较了两者的收敛性和分布性,证明由克隆进化算法得到的结果距离真实Pareto曲线更接近,分布更均匀、范围更广泛。

Both the clonal selection principle and biological evolution theory are applied in the multi-objective optimization. It uses memory eell of immunity clone theory in this algorithm, which uses the cluster method in the memory cell set of the clonal selection algorithm to renew and eliminate antibody. The non-uniform mutation operator is employed to the multiplicity of population. This algorithm promotes to individual competition by antibody-antibody affinity. Populations get universality and restrain excessive competition with antigen-antigen affinity. Final results show that the proposed approach have performances similar or better than those produced by NSGA-Ⅱ. It get a closer Pareto curve which is a set of the uniform and widespread solution.