针对标准粒子群优化算法在求解复杂多模问题时容易陷入局部极值点和有限冲击响应数字滤波器(FIR DF)设计时减少误差的问题,将综合学习粒子群优化算法(CLPSO)应用于FIR DF设计中.CLPSO在每一代更新中采用所有粒子全局最优值代替粒子本身的个体历史最优值,当粒子停止更新时,重置粒子最优值,保证粒子学习最优和在错误方向上花费最少计算时间.数值结果显示,在满足算法复杂度、计算时间、逼近误差等设计指标的前提下,CLPSO在低通和高通频率采样法FIR DF设计中比传统查表法、遗传算法和标准粒子群优化算法具有一定的优势.%Because standard particle swarm optimization is prone to fall into local extreme points in solving complex multi-mode, and for reducing the error in the digital filter design of finite impulse response (FIR DF) , the comprehensive learning particle swarm optimization algorithm is applied to designing FIR DF. In each updating generation, the global optimum values of all particles are taken instead of the individual history optimal one. When the particle updating stops, the optimal value of the particle gets reset to ensure particle to learn best with shortest computing period in wrong directions. The numerical results show the advantages in the low-pass and high-pass frequency sampling FIR filter design, over the traditional look-up table method, genetic algorithm and standard particle swarm optimization algorithm for the same design requirements, such as the calculating task and algorithm complexity.
展开▼