基于多分辨分析和WNN的模拟电路故障诊断

提出了一种基于多分辨分析和小波神经网络(WNN)相结合的模拟电路故障诊断方法。该方法利用了多分辨分析优异的时频特性,提取采集数据中的故障特征参数值,结合小波神经网络强大的非线性分类、学习、泛化能力及精度高、收敛速度快等特性,将得到的输入数据进行归一化处理作为小波神经网络的输入对其进行训练,并将训练的结果应用于滤波器电路故障诊断。结果表明,该方法实现了对故障模块的定位,是一种有效的模拟电路故障诊断方法。

Analog circuit fault diagnosis based on the multiresolution analysis and wavelet neuralnetwork

A method of analog circuit fault diagnosis based on the multiresolution analysis and wavelet neural network（WNN）is proposed. The novel method makes full use of advantages of time-frequency localization of multiresolution to extractthe feature parameter values from the collected date. In combination with strong ability of nonlinear classification,learning,generalization,high accuracy and fast convergence of WNN,the normalization processing of the obtained input data is taken asthe inputs of WNN to complete its training. Finally the training result is applied to the fault diagnosis of the filter circuit. The re-sult shows that the method can realize accurate localization of the fault module,and is a effective means for fault diagnosis ofanalog circuit.