基于小波包特征提取的ART1网络故障诊断研究

系统阐述了小波包分解及ART1神经网络的基本理论和特点,在对故障诊断进行理论分析的基础上,结合ART1神经网络对输入矢量的要求,提出了一种差值门限方法,将小波包特征提取与ART1神经网络相结合进行动态系统的故障诊断。用MATLAB仿真了系统及其故障,对上述故障诊断方法进行了测试实验。实验结果表明,基于小波包特征提取的ART1神经网络故障诊断方法是可行的,能够应用于多个领域,其发展前景广阔。

Fault Diagnosis Based on Wavelet Packets and ART1 Neural Networks

The paper describes the theory and the characteristics of the wavelet-packets and the ART1 Neural Networks.On the basis of the theoretical analysis of fault diagnosis and the input requirement of the ART1 NN,a method is presented,which uses the D-value threshold integrating the feature extraction by the wavelet-packets and the ART1 NN to diagnose the dynamic system failure.Simulation of the system and its faults is done to test the method,the result indicates that the method is workable,versatile and promising.