用神经网络建立非线性系统模型研究

Research on Modelling Non-Linear Systems Using Neural Networks

中文关键词  多层神经网络  辨识  递推预报误差法  非线性系统  建模  PWM直流调速系统

英文关键词  multilayered neural networks  identification  recursive prediction error algorithm  non-linear system  model  PWM speed control system

中文摘要

      本文针对多层神经网络结构，运用递推预报误差（RPE）算法对离散非线性系统进行辨识研究。作为运用实例，本文对一个工业实际对象进行了神经网络动态建模。研究结果表明，神经网络方法是用于带有非线性特性工业过程建模的有效方法。

英文摘要

      This paper discusses the use of recursive prediction error algorithm of training multilayered neural networks for discrete time non-linear system identification. Results based on the simulated system and the dynamic modelling of an industrial example indicate that the recursive prediction error algorithm is attractive and effective for modelling of complicated industrial processes which possess non-linear characteristics.