基于神经网络非线性系统的故障诊断研究

An approach to fault diagnosis of nonlinear system based on neural networks

提出了一种关于变工作点非线性系统故障诊断研究的新方法。将此类非线性系统用变参数线性系统表示,其中模型参数为可测量工作点及故障的函数。基于Hopfield神经网络,估计系统模型参数,引入“参考工作点”这一新概念,根据模糊聚类分析方法,确定故障种类。最后,在某位置伺服系统故障诊断研究中证实了这种方法的有效性。

A new approach to fault diagnosis of a nonlinear system whose operating point always varies is presented.This kind of nonlinear system is modeled by linear system with varying parameters,the model parameters are functions of the measured operating points and the faults,they can be evaluated based on Hopfield Neural Networks.The reference operating point is introduced.The fault can be determined based on fuzzy cluster analysis.Finally the effectiveness of this method is verified in a position servo system