基于RBF神经网络的非线性系统故障诊断

针对一类模型未知的非线性动态系统,提出了一种基于神经网络在线估计结构的鲁棒故障诊断检测方法.系统中,仅输入输出可测,且包含输出不确定性项.该方法通过构造神经网络在线逼近结构来拟合该非线性系统模型和系统的非线性故障特性,建立系统的状态观测器.神经网络的权重调整规律由李亚普诺夫稳定性方法获得,系统的输出不确定性部分被用于获得诊断算法的阈值.最后,用Matlab/SIMULINK对的算法予以验证,结果表明本算法的合理性.

Fault Diagnosis of Nonlinear System Based on RBF

A robust fault diagnosis method based on neural networks on-line approximation structure for a class of nonlinear systems with unknown model is proposed. The only measurable variables are the inputs and outputs of the system. There are uncertain parts in the output. A nonlinear on-line approximator based on RBF is utilized to estimate the unknown model in the nonlinear system, simultaneously, to monitor the faults and estimate the fault value, and present a state observer for the nonlinear system. The method of turning the weights of neural networks is got by Lyapunov stability theory. The uncertain parts are used to get the deadline of the fault diagnosis. At last, a simulation example illustrates the effectiveness of the approach by MATLAB/SIMULINK.