摘要
针对滚动轴承故障振动信号的非平稳特征,提出一种基于总体经验模态分解(EEMD)和模糊BP神经网络的故障诊断方法。首先对滚动轴承的振动信号采用总体经验模态分解方法进行分解,得到若干个本征模态函数分量(IMF);然后提取各分量的均方差、峭度和能量,把这些特征参数作为学习集和训练集,将学习集输入到模糊BP神经网络中进行学习;最后把训练集输入到特征参数经过学习训练后的模糊BP神经网络中进行故障类型识别,并与BP神经网络进行比较。实验结果表明:所提方法能有效地应用于滚动轴承故障诊断,而且比BP神经网络具有更高的精确度。
Considering non-stationary characteristics of bearing＇s fault vibration signals,an EEMD and fuzzy neural network-based fault diagnosis method was proposed,in which,having EEMD method adopted to decompose the bearing＇s bearing vibration signals into several intrinsic mode function components（IMF）; and then,extracting each component＇s mean square error,kurtosis and energy and taking them as learning set and training set; and finally,having the learning set input into the fuzzy BP neural network for learning and the training set into the fuzzy BP neural network for fault type identification. Experimental results show that,the method proposed can be effectively applied to the rolling bearing＇s fault diagnosis and it outperforms the BP neural network in the accuracy.
出处
《化工自动化及仪表》
CAS
2017年第1期34-38,72,共6页
Control and Instruments in Chemical Industry
基金
国家自然科学基金项目(61263023)
