论文部分内容阅读
作为信号处理领域的重要研究课题 ,滤波器设计本质上是一个多维参数寻优问题 ,且往往存在多极小。传统最小二乘法和单纯形法易陷入局部极小 ,而单一模拟退火算法搜索过程冗长 ,单一遗传算法易早熟收敛。结合模拟退火的随机概率突跳性搜索和单纯形法的凸多面体几何搜索 ,提出了有效设计自适应IIR滤波器的一种简单易实现的单纯形 退火策略 (simplexmethod simulatedannealing ,SMSA) ,并给出了算法操作和参数的合理设计方案。基于多个典型系统的随机数值仿真以及与最小二乘方法的比较研究 ,验证了所提方法的有效性、全局优化性和初值鲁棒性。
As an important research topic in the field of signal processing, the filter design is essentially a multi-dimensional parameter optimization problem, and often there are many minima. The traditional least square method and simplex method are trapped in local minima, while the single simulated annealing algorithm has a long searching process and the single genetic algorithm tends to converge prematurely. Combined with the random probabilistic jump search of simulated annealing and convex polyhedron geometric search of simplex method, a simple and easy simplex simulated annealing (SMSA) method is proposed to effectively design adaptive IIR filter. A reasonable design of algorithm operation and parameters. Based on the random numerical simulation of several typical systems and the comparison with the least square method, the validity, global optimization and initial robustness of the proposed method are verified.