基于小波神经网络的模拟电路故障诊断改进

在运用神经网络进行模拟电路故障诊断的过程中,代表着故障特征的网络输入至关重要,由于小波变换的时频局部化和多尺度分析等特性,将两者结合起来,通过小波变换对模拟电路的输出响应进行故障特征提取,同时解决PSPICE与MATLAB之间的数据通信问题,提出将蒙特卡罗分析产生的所有训练样本经过处理后输入到一个神经网络进行训练的方法,从而避免了训练多个神经网络。利用神经网络对各种故障模式进行分类,实现模拟电路的故障诊断,并进一步与传统的BP网络故障诊断法进行比较。仿真结果表明,该方法可以实现故障检测及定位,诊断的准确率显著提高,适用于模拟电路故障诊断。

Improvement methods for fault diagnosis of analog circuits based on wavelet neural networks

In the process of using neural network for fault diagnosis of analog circuits,the network input which represents fault signature is very important.Combining the two features of time-frequency location and multiple-scale analysis of wavelet transform,obtain the fault features of analog circuit out put by wavelet transform,So will both together and the collected data was preprocessed by wavelet transform to generate,and also solves the data communication problems between PSPICE and MATLAB.A method is proposed that input all treatment training samples which are produced by Monte-Carlo analysis to one neural network,which avoids training multiple neural networks.Then realized fault diagnosis of analog circuits by neural network to classify the various fault modes.Compared with the traditional method which only uses BP neural network to diagnose,simulation results show that this method is propitious for fault diagnosis of analog circuits which can realize fault detection and localization,and accuracy rate of diagnosis is improved significantly.