基于未知输入观测器的不确定非线性系统故障检测

针对具有不确定性和系统故障的非线性系统，利用神经网络构造了全阶未知输入观测器，在获得系统的状态观测信号的同时得到了系统的故障观测信号。通过在故障观测神经网络权值的调整规律中引入死区函数，从而提高了故障观测对系统不确定性的鲁棒性。利用得到的故障观测信号，可以方便地检测系统的缓变故障和突变故障，实现了对系统故障的快速检测，降低了误检率。仿真示例表明了该方法的有效性。

Fault Detection for Uncertain Nonlinear System Based on Unknown Input Observer

For a class of uncertain nonlinear systems with an unknown input, a new full-order unknown input observer is developed. Using the proposed method, the state and the fault can be observed at the same time. A dead zone function is used for the adjust rule of the weight matrix of the fault observer neural network, so that the robust of the fault is improved under the uncertainty of the system. In this method, the fast abrupt fault and the slow variation fault of the system are conveniently detected, and the detection error rate is reduced. Simulations show the effectiveness of the proposed method.