基于非线性系统的自适应神经网络均衡器特性分析  

Characteristic Analysis of Adaptive Neural Network Equalizer Based on Nonlinear System

作　　者：任志勇[1] REN Zhiyong(School of Communication Engineering Chongqing College of Electronic Engineering,Chongqing 401331,China)

机构地区：[1]重庆电子工程职业学院通信工程学院,重庆401331

出　　处：《电工技术》2022年第9期32-33,36,共3页Electric Engineering

基　　金：重庆市科技计划项目(编号cstc2019jscx-msxmX0098)。

摘　　要：针对非线性系统中的时变及参数动态变化引起的非线性失真及频谱衰落,提出了一种自适应神经网络均衡器结构与反向传播的遗传学习算法。均衡器前馈部分采用RBF神经网络,用于对非线性系统信道的逼近,而反馈部分采用基于递归最小二乘算法的判决反馈。实验验证了该判决反馈的自适应神经网络均衡器能有效消除非线性系统信道干扰,在误比特率方面改善了均衡器的性能。For nonlinear distortion and spectrum fading caused by time-varying and parameter dynamic changes in nonlinear systems,an adaptive neural network equalizer structure and back-propagation genetic learning algorithm are proposed.The feedforward part of the equalizer adopts the RBF neural network,which is used to approximate the nonlinear system channel,and the feedback part adopts the decision feedback based on the recursive least squares algorithm.Experiments show that the neural network adaptive equalizer with decision feedback can effectively eliminate the channel interference of nonlinear systems,and improve the performance of the equalizer in terms of bit error rate.

关 键 词：非线性系统 自适应均衡器 频谱衰落 神经网络 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]