摘要
为提高类电磁机制算法的优化性能,提出了一种改进的类电磁机制算法.引入早熟的判断和处理机制,对陷入早熟状态的带电粒子,迭代时以一定的概率选中进行小波变异扰动,以提高算法的种群多样性,从而克服EM算法后期易陷入局部最优的缺陷.数值仿真结果表明:改进后的算法搜索精度提升、稳定性更好.该算法用于设计FIR低通数字滤波器和带通数字滤波器取得了较为满意的结果.
Electromagnetism-like mechanism algorithm is difficult to deal with premature and local convergence.In order to solve this problem,an improved EM algorithm is proposed.In the proposed algorithm,mutation operator is adopted by selecting particles with certain small probability to improve population diversity,so as to overcome the EM′s drawback of occurring premature convergence and trapping in the local optima.Experimental results on benchmark functions show higher accuracy and stability of the new algorithm.Besides,the algorithm is applied to the low pass and band pass FIR digital filters with more satisfactory results.
作者
胡晓朋
田雨波
许永秀
李双双
HU Xiaopeng;TIAN Yubo;XU Yongxiu;LI Shuangshuang(School of Electronics and Information,Jiangsu University of Science and Technology,Zhenjiang 212003,China)
出处
《江苏科技大学学报（自然科学版）》
CAS
2018年第4期549-554,共6页
Journal of Jiangsu University of Science and Technology:Natural Science Edition
基金
江苏省重点研发计划社会发展基金资助项目(BE2016723)
关键词
类电磁机制算法
小波变异
FIR数字滤波器
electromagnetism-like mechanism algorithm
wavelet mutation
FIR digital filters
