神经网络自适应滤波器

Adaptive Neural Network Filters

本文介绍一种新颖的非线性自逅应滤波器——自适应神经网络滤波器.由于神经网络具有学习非线性函数到任意的精度以及自适应能力,这种滤波器优于线性滤波器,能适应各种噪声环境.在自适应LMS算法基础上,提出了在线PB训练算法、收敛速度快.最后以自适应噪声对消系统为例,进行了计算机仿真,结果显示了这种滤波器的良好性能.

This paper describes a novel type of nonlinear adaptive filters——adaptive neural net-work filters. Since the neural network is capable of learning a nonlinear function to any arbi-trany degree of accuracy and adaptation, this type of filter performs better than that of linear filter and can be adapted to different kinds of noisy environments. Based on adaptive LMS algorithms, a BP algorithm on line with fast convergence speed is presented. Finally,an adaptive noise cancelling system is designed as an example and the computer simulation results show the superior performance of the adaptive neural network filters.