进化论自适应滤波算法及降噪特性研究

基于生物进化论策略的自适应滤波算法，利用其有性繁殖和无性繁殖的随机搜索能力实现全局最优搜索，可克服最小均方法（Least Mean Square，简称：LMS）对于多峰特性问题有可能收敛于某一局部最小值，无法实现全局最优的缺点。提出峰值系数指标PR（Peak Ratio）概念解决进化论自适应滤波器的性能评估问题，提出收敛速度CS（Converge Speed）指标和峰值系数指标PR高度的方法解决最优滤波参数的选取问题，并列进化论自适应滤波器的降噪进行模拟仿真，讨论了滤波长度和进化系数对进化论自适应算法消噪效果的影响及最优滤波参数。

Noise Cancellation Using an Evolutionary Adaptive Algorithm

The least mean square (LMS) algorithm popularly used in adaptive filtering design is based on a gradient search algorithm, it typically converges to local minima when solving multiple-peak problems. The Evolutionary Digital Filter (EDF) , which uses the concept of evolutionary strategies of living things, combined with cloning methods and mating methods, can overcome the shortcomings of the gradient search algorithm in solving multiple-peak problems. The concept of Peak Ratio ( PR) is introduced for evaluating the behavior of EDF, and the indices of Converge Speed ( CS) and Peak Ratio ( PR) are proposed for selecting the optimal data length. Simulation results show that the EDF using the evolutionary algorithm is an effective method for noise cancellation. The influences of the data length and evolutionary factor for the EDF are discussed.