基于自适应神经网络的非线性系统故障诊断

针对一类模型未知及状态不可测的非线性系统,提出了基于自适应神经网络的故障诊断策略,不仅在线估计神经网络的矩阵权重,而且在线估计高斯函数的宽度和中心。该方法对系统的未知非线性特性没有特别要求,仅对神经网络提出较弱的假设条件。首先利用径向基函数(Radial Basis Function,简称RBF)神经网络构造状态观测器,估计系统的状态。然后利用另一个自适应RBF神经网络作为故障估计器,其输入是系统的估计状态(而不是系统状态),其输出为系统所发生的故障模型。利用Lyapunov稳定理论详细分析了状态误差和故障误差的收敛性,分别给出了两个神经网络的参数调整律,仿真证明了该方法的实用性和有效性。

Fault diagnosis based on adaptive nueral networks for a class of nonlinear systems

Fault diagnosis architecture based on adaptive neural networks for a class of unknown nonlinear systems with unmeasured states is proposed.The center vector and width vector of Gaussian function are on-line updated but updating weight matrix.Under the mild condition,the problem of fault diagnosis can be solved for the nonlinear systems.The states of system and the faults in system are estimated respectively by employing two RBF neural networks.Estimated states are input to the fault approximator whose outputs are estimated fault.The stability of the error system is analysized in detail,the parameter updating laws for two neural networks are given.Finally,a simulation example is given to illustrate the effectiveness of the approach.