机构地区: 武汉理工大学计算机科学与技术学院
出 处: 《武汉理工大学学报(信息与管理工程版)》 2005年第2期192-196,共5页
摘 要: 提出了一种新型的PSO变异策略———CPg变异,该变异策略的首先定义了全局收敛度最大位置C,并在搜索循环的每次迭代中,以一定的概率交替使用C和Pg来代替原迭代公式中的Pg。通过对4个多峰的测试函数所做的对比实验表明,CPg变异增强了搜索能力,求得全局最优的成功率和收敛到速度大为提高,克服了原始的PSO算法易于收敛到局部最优点的缺点,也明显优于对原始PSO进行传统变异的方法。 A new type of PSO mutation strategy namely CP g mutation is presented. The position C of maximum convergence is defined. In each iteration of the searching cycle, the iteration formula is replaced by C and P g alternately according to certain probability. Through the comparison test among four multi-peek functions, it is demonstrated that the CP g mutation enforces the searching capability, the successful searching rate is greatly increased, and the shortcoming of original PSO algorithm's liability to convergence to a local optimum is avoided. It is also superior to PSO traditional mutating methods.
领 域: [自动化与计算机技术] [自动化与计算机技术]