一类非线性系统的多故障检测方法研究

Study on Multiple-fault Detecting Method for Non-linear System

针对一类非线性系统,提出了一种基于最优滤波的多故障检测观测器设计方法。引入神经网络逼近系统非线性部分,由Lyapunov稳定性理论设计神经网络权值调整规则.采用对等空间方法设计最优滤波矩阵滤除干扰故障噪声,从而对目标故障进行检测和跟踪。仿真结果验证了该方法的正确性和可行性。

A novel multiple-faults diagnosis method for non-linear system was proposed. The nonlinear part was approximated by neural network, and the on-line tuning rule of neural network was designed by Lyapunov stability theory. Parity space approach was introduced to design the optimal filter matrix, which used to filter out the nuisance faults, so the target fault could be detected and tracked. Finally, the simulation example shows the validity and the applicability of the proposed method.