An expert system for fault diagnosis in internal combustion engines using wavelet packet transform and neural network
An expert system for fault diagnosis in internal combustion engines using wavelet packet transform and neural network
复制标题
DOI：
10.1016/j.eswa.2008.03.008
复制
发表时间：
2009-04-01
影响因子：
8.5
通讯作者：
Liu, Chiu-Hong
中科院分区：
文献类型：
--
作者：
Wu, Jian-Da;Liu, Chiu-Hong
本文提出了一种基于小波包变换和人工神经网络的内燃机故障诊断系统。在机械系统的故障诊断中，小波变换是一种众所周知的用于故障检测和识别的信号处理技术。本系统的信号处理算法借鉴了前人在语音识别方面的研究成果。在声发射信号的预处理中，利用小波变换系数来评价声发射信号的熵值，并将其作为识别故障状态的特征。显然，小波变换可以改善连续小波变换（CWT）较长的计算时间和较大的操作数。它还可以通过离散小波变换（DWT）分解近似版本来解决频带不一致问题。在实验中，这些小波被用作母小波来构建和执行所提出的小波变换技术。在分类中，为了验证广义回归神经网络（GRNN）在故障诊断中的效果，将传统的反向传播网络（BPN）与广义回归神经网络（GRNN）进行了比较。实验结果表明，该系统在不同工况下的平均分类准确率均在95%以上。(c) 2008 Elsevier Ltd.版权所有。
In the present study, a fault diagnosis system is proposed for internal combustion engines using wavelet packet transform (WPT) and artificial neural network (ANN) techniques. In fault diagnosis for mechanical systems, WPT is a well-known signal processing technique for fault detection and identification. The signal processing algorithm of the present system is gained from previous work used for speech recognition. In the preprocessing of sound emission signals, WPT coefficients are used for evaluating their entropy and treated as the features to distinguish the fault conditions. Obviously, WPT can improve the continuous wavelet transform (CWT) used over a longer computing time and huge operand. It can also solve the frequency-band disagreement by discrete wavelet transform (DWT) only breaking up the approximation version. In the experimental work, the wavelets are used as mother wavelets to build and perform the proposed WPT technique. In the classification, to verify the effect of the proposed generalized regression neural network (GRNN) in fault diagnosis, a conventional back-propagation network (BPN) is compared with a GRNN network. The experimental results showed the proposed system achieved an average classification accuracy of over 95% for various engine working conditions. (c) 2008 Elsevier Ltd. All rights reserved.