基于免疫克隆原理的改进粒子群优化算法的研究

提出了一种改进的粒子群优化(PSO)算法来进行函数优化,以克服PSO算法容易陷入局部极值的不足,加快收敛速度,从而实现全局搜索.PSO算法是基于群体智能的随机优化算法,参数结构简单,但收敛速度慢,容易陷入局部极值.通过对PSO算法的深入分析,基于传统的速度--位置更新操作,把免疫克隆(IC)原理引入PSO算法中,将抗体视为粒子,根据亲和度的高低进行粒子克隆选择、克隆抑制和高频变异,提高了种群的多样性和全局搜索的能力.测试结果表明,该算法完成全局搜索所需的迭代次数明显少于PSO算法,大大缩短了搜索时间,在多维函数最优解的搜索中具有优良的性能.

Study of Modified Particle Swarm Optimization Algorithm Based on Immune Clone Principle

A modified particle swarm optimization (PSO) algorithm was adopted to optimize functions, which overcame the shortcoming of converging to local optimum for PSO algorithm, increased the converging rate and achieve the global searching. PSO algorithm is a random optimizing algorithm based on swarm intelligence that has a simple parameter structure; however, it has a slow converging rate and is easy to obtain a local optimum. Through a considerate analysis of PSO algorithm, immunity clone (IC) algorithm was introduced to the PSO algorithm based on traditional velocity-displacement operator. The antibodies could be regarded as the particles, and according to the degree of affinity, the clone selection, clone suppression, and high-frequency mutation were performed, which could enhance the diversity of particle swarms and the capability of global searching. From the test results, it is shown that this algorithm has perfect property in multi-dimension function searching and needs shorter searching time and fewer iteration times than PSO algorithm.