摘要
神经网络具有大规模并行处理、自适应、自学习能力和非线性特征，已获得了广泛应用。本文将多层正馈神经网络（ＭＬＦＮＮ）应用到数字通信中的均衡问题，讨论了ＭＬＦＮＮ自适应均衡（ＡＥ）和判决回授均衡（ＤＦＥ）的结构及其学习算法。通过计算机仿真研究了ＭＬＦＮＮ组成的ＡＥ和ＤＦＥ的性能以及系统收敛情况。初步结果表明，神经网络用于通信中的均衡器是可行的、有效的，性能是优良的。
This paper discusses the applicationer of the multilay feedforward neural network(MLFNN)to the equalization problem in digital communication and analyses the struc-tures,convergence characteristics and performance. The results from the com putersimulation indicate that the application of the neural net to the equalizer used in communi-cation is feasible,effective and excellent in performance.
关键词
神经网络
自适应均衡器
数据通信
neural network, BP training algorithm,adaptive equalization,bit errorrate
