非线性系统的在线鲁棒故障检测

提出了一种具有建模不确定性的非线性系统在线故障检测方法。故障被假定为状态变量和输入变量的函数，该系统仅是输入、输出可测量的。一种基于RBF神经网络的在线非线性估计器用来跟踪系统中的出现的故障，该估计器对有建模不确定性的非线性系统的故障检测具备良好的鲁棒性。文中所提出的方法的收敛性在理论上进行了较详细证明。仿真实例说明了该故障检测方法的有效性和实用性。

On-Line Robust Fault Detection in a Class of Nonlinear Systems

A kind of on-line fault detection scheme in nonlinea r systems with modeling uncertainty is presented. The faults are assumed t o be a function of the state and the input. Only the inputs and the outputs of s ystem can be measured. An on-line approximator based on RBF neural network is u sed to track the faults of the system, and the approximator has good robustness for the fault detection in the nonlinear system with modeling uncertainty. The c onvergence of the scheme derived in the paper is proved in the theory. An exampl e demonstrates the efficiency and the applicability of the fault detection sche me.