基于小波神经网络的多模型故障检测

提出了一种利用小波神经网络辨识非线性系统多模型故障的方法。证明了状态估计误差渐近收敛到零，同时证明了如果激活函数满足持续激励条件，辨识器参数将趋于理想辨识器参数。分析了多模型辨识结构，并将小波神经网络作为辨识器应用于多模型故障检测。给出了小波神经网络进行非线性系统逼近的实例，用小波神经网络辨识器对多故障模型检测进行了仿真，证明了此方法的正确性和可行性。

Multiple Model Failure Detection Based on Wavelet Neural Network

The method for the multiple model failure detection is presented based on wavelet neural network and the designed neural network observer to increase the precision of the identification. Meanwhile the state estimation error is proved to be converged to zero asymptotically, and parameters of the identifier are converged to the ideal identifier parameters if persistency of excitation condition of the activation function is fulfilled. The wavelet neural network as the multiple model identification structure is analyzed and applied to the multiple model failure detection. Numerical example that the wavelet neural network approximated the nonlinear system is described, and the simulation of multiple model failure detection using wavelet neural network identifiers is introduced while the noise disturbance is added to the system, and the correctness and the feasibility of the detection method are illustrated.