基于神经网络的Wiener系统故障检测方法  

Fault detection study of Wiener systems using neural networks

作　　者：沈艳霞[1] 朱嵘嘉[1] 纪志成[1] 

机构地区：[1]江南大学控制科学与工程研究中心,无锡214122

出　　处：《东南大学学报（自然科学版）》2005年第A02期95-99,共5页Journal of Southeast University：Natural Science Edition

基　　金：江苏省高技术研究计划资助项目(BG2005014)

摘　　要：针对一类线性部分严格正定且非线性部分可逆的Wiener模型,提出了基于神经网络参数估计的系统线性部分故障检测方法.引入卡尔曼滤波器滤除系统未知干扰和噪声,并将Wiener模型转化为ARMA模型,计算得到线性部分参数;神经网络在线估计系统线性部分的参数,通过参数估计值与实际值的比较检测系统故障.仿真实例证明了该方法的正确性和有效性.On conditions that the linear element be positive define and the non-linear part be invertible, a novel fault detection method for Wiener system is presented. The neural networks are adopted to estimate the parameters. The Kalman filter was introduced to overcome the unknown noise. By transforming the Wiener model to ARMA model, linear element parameters can be obtained. With the neural networks estimating values, the fault can be detected by real time comparing parameters estimation values with real values. Finally, a simulation example is given to show the validity and the applicability of the proposed approach.

关 键 词：WIENER模型 卡尔曼滤波算法 神经网络 参数估计 故障检测 

分 类 号：TP118[自动化与计算机技术—控制理论与控制工程]