摘要
利用小波与改进算法的BP神经网络相结合的方法进行模拟电路故障诊断,该方法使用小波分解作为预处理工具,对信号进行消噪和小波分解,然后提取特征信息,进行归一化处理,并作为BP神经网络的输入样本进行模式识别。该方法减少了神经网络的输入维数,提高了收敛速度和辨识故障的能力。仿真结果表明,该方法能准确快速地定位故障,且可有效地进行故障识别、改善神经网络结构以及提高故障诊断精度与速度。
This paper presents a systematic fault diagnosis method for analogue circuits based on the combina- tion of wavelet transform and modified BP neural network （BPNN）. With the wavelet decomposition as a preprocessor, the feature information is extracted by wavelet denoising and normalization and applied to the BPNN. The method reduces the number of inputs to the neural network and improves the convergence speed and the ability to detect faults. Simulation results show that the scheme locates faults exactly with improved network structure and fault diagnosis precision and speed.
出处
《电子科技》
2014年第8期46-49,共4页
Electronic Science and Technology
关键词
小波变换
BP网络
模拟电路
故障诊断
wavelet transform
BP networks
analogue circuits
fault diagnosis
