基于遗传算法的自适应噪声抵消  

Adaptive noise cancellation based on genetic algorithm

作　　者：郑陶冶[1] 高翔[1] 

机构地区：[1]东南大学无线电工程系,南京210096

出　　处：《声学技术》2003年第1期26-29,共4页Technical Acoustics

基　　金：江苏省应用基础研究资助项目(BJ98002)

摘　　要：当有参考噪声信号时,自适应噪声抵消的实质就是求参考噪声输入通路的逆滤波器,LMS自适应滤波问题就是一个多变量函数的极值问题。LMS算法因其具有算法简单、容易实现的优点而为常用,但是算法的收敛特性和失调量受到步长参数μ的影响,而步长参数μ的最优值不易确定。遗传算法是一种应用于大规模搜索空间的有效方法,它不要求函数的解析表达式,只根据已知的测量数据便可以求得全局极值。本文以FIR滤波器为例,采用改进的实值编码遗传算法,将遗传算法用于逆滤波器的求解。计算机仿真结果表明该算法对噪声抵消取得了较满意的效果。The essence of adaptive noise cancellation is to solve an inverse filter of the reference channel when the reference signal is available, and the least mean square (LMS) adaptive filtering algorithm is to seek the extreme values of a multivariable function. The LMS algorithm is commonly used due to its simplicity and ease of implementation. However, the convergence behavior and misadjustment of the algorithm is seriously affected by the stepsize μ, and the optimum value of μ cannot be easily determined. Genetic algorithm is an effective method applicable to largescale searching space. Even if the function is unknown, GA can still produce the global optimum from measured data. Taking FIR filter as an example, this paper uses an improved realcoded genetic algorithm to solve the inverse filter problem. Simulation shows that this method can give satisfactory results.

关 键 词：遗传算法 自适应滤波 自适应噪声抵消 

分 类 号：TN713[电子电信—电路与系统]