一种推广形式的自适应算法
A Family of Generalized Adaptive Algorithm
-
摘要: 在某些实际问题中,如地震勘探信号处理等,对于非最小相位系统相位特性辨识,及非正 态分布随机信号的处理问题,用非二次范数为准则可以得到比传统的最小二乘法更好的结果, 但这常导致解一组非线性方程使计算复杂.本文提出一种新的最陡梯度下降自适应算法,它 是LMS法的推广,计算简单,可以误差的任意次范数为目标函数.文章证明了算法的收敛性 及其收敛结果,并给出了实验结果.Abstract: Recently, in seismic signal processing, problems of identification of phase characteristics for non-minimum phase system and processing of non-Gaussian distributed random signal have been posed. Some research results have shown that it is advantageous to use nonmean square criterion in stead of the traditional I,S method. But this often leads to solving some non-linear equations, which needs tedious computation. A new family of steepest descent algorithm for adaptive processing has been proposed in this paper. It is a generalization of the traditional LMS algorithm and allows error minimization in arbitrary norm. Convergence by means of this new algorithm has been derived, and experimental results also given.
点击查看大图
计量
- 文章访问数: 1384
- HTML全文浏览量: 58
- PDF下载量: 1276
- 被引次数: 0