摘要
针对滚动轴承故障振动信号的非平稳特征,提出了基于经验模态分解(Empiri calModeDecomposition,简称EMD)和神经网络的滚动轴承故障诊断方法.首先对原始信号进行了经验模态分解,将其分解为多个平稳的固有模态函数(IntrinsicModefunction,简称IMF)之和,再选取若干个包含主要故障信息的IMF分量,并从中提取时域特征指标———峭度或裕度因子作为神经网络的输入参数来识别滚动轴承的故障模式.对滚动轴承的内圈、外圈故障信号的分析结果表明,以EMD为预处理器提取时域特征参数的神经网络诊断方法比直接从原信号中提取时域特征参数的诊断方法有更高的故障识别率,可以准确、有效地识别滚动轴承的工作状态和故障类别.
According to the nonstationary characteristics of roller bearing fault vibration signals, roller bearing fault diagnosis method based on Empirical Mode Decomposition (EMD) and neural networks is put forward in this paper. First of all, original signals are decomposed into a finite number of stationary Intrinsic Mode functions (IMFs), then a number of IMFs containing main fault information are selected, from which time domain feature indicator-kurtosis or allowance which serves as input parameter of neural networks is extracted in order to identify fault patterns of roller bearing. The analysis results from roller bearing signals with innerrace and outrace faults show that the approach of neural network diagnosis based on EMD extracting time domain features is superior to the method which extract time domain feature from original signals directly and would identify roller bearing fault patterns accurately and effectively.
出处
《湖南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2003年第5期25-28,共4页
Journal of Hunan University:Natural Sciences
基金
国家自然科学基金(50275050)
高等学校博士点专项科研基金(20020532024)资助项目
