基于小波包分解与局部均值分解排列熵的自适应轴承故障诊断
|
摘要
轴承运行过程是一个复杂的非平稳动态过程,提出小波包分解与局部均值分解排列熵相结合的特征提取方法,以支持向量机为故障模式识别器,对轴承故障进行诊断。首先对原始振动信号进行小波包阈值消噪处理,根据特征频率进行频带划分及信号重构;然后采用局部均值分解方法将重构信号自适应分解为若干模态分量,并计算包含主要故障信息分量的排列熵,实现对模态分量的特征量化;最后将熵值特征向量输入多分类支持向量机,用于判断轴承的故障类型及故障程度。分析结果表明,这一方法的轴承故障诊断识别率可达95%,与其它方法相比,这一方法能够有效提取轴承故障特征,具有更高的识别准确率。
Bearing run procedure is a complex non-stationary dynamic process. A feature extraction method combining WPD and LMD permutation entropy was proposed to diagnose the bearing fault when the SVM was used as fault mode recognizer. Firstly, the wavelet packet threshold de-noising was performed on the original vibration signal and the band division and signal reconstruction were carried out according to the characteristic frequency. Then the LMD method was adopted to decompose the reconstructed signal into several modal components, and the permutation entropy including the main fault information component was calculated to realize the feature quantization of the modal component. Finally, the feature vector of the entropy value was input into the multi-classification SVM in order to estimate the fault type and fault degree of the bearing. The analytic results show that this method can achieve 95% bearing fault recognition during diagnosis. Compared with other methods, this method can effectively extract the bearing fault signature with higher recognition accuracy.
Bearing run procedure is a complex non-stationary dynamic process. A feature extraction method combining WPD and LMD permutation entropy was proposed to diagnose the bearing fault when the SVM was used as fault mode recognizer. Firstly, the wavelet packet threshold de-noising was performed on the original vibration signal and the band division and signal reconstruction were carried out according to the characteristic frequency. Then the LMD method was adopted to decompose the reconstructed signal into several modal components, and the permutation entropy including the main fault information component was calculated to realize the feature quantization of the modal component. Finally, the feature vector of the entropy value was input into the multi-classification SVM in order to estimate the fault type and fault degree of the bearing. The analytic results show that this method can achieve 95% bearing fault recognition during diagnosis. Compared with other methods, this method can effectively extract the bearing fault signature with higher recognition accuracy.
引文
[1]李泽豪,顾海明,张亿雄.基于小波包和EMD的滚动轴承故障诊断[J].煤矿机械,2010,31(6):243-245.
[2]MA J,WU J D,FAN Y G,et al.The Rolling Bearing Fault Feature Extraction Based on the LMD and Envelope Demodulation[J]Mathematical Problems in Engineering,2015,2015(6):1-13.
[3]TIAN Y,WANG Z L,LU C.Self-adaptive Bearing Fault Diagnosis Based on Permutation Entropy and Manifold-based Dynamic Time Warping[J/OL].http://doi.org/10.1016/j.ymssp.2016.04.028.
[4]程军圣,史美丽,杨宇.基于LMD与神经网络的滚动轴承故障诊断方法[J].振动与冲击,2010,29(8):141-144.
[5]颜天晓,张瑞亮,王铁,等,基于EEMD和Hilbert包络分析的轴承复合故障诊断研究[J].机械传动,2016(6):132-135.
[6]杨斌,程军圣.基于WPD-LMD和排列熵的结构损伤识别方法[J].湖南大学学报(自然科学版),2014,41(8):41-46.
[7]冯辅周,饶国强,司爱威,等.排列熵算法研究及其在振动信号突变检测中的应用[J].振动工程学报,2012,25(2):221-224.[J].
[8]郑近德,程军圣,杨宇.基于LCD和排列熵的滚动轴承故障诊断[J].振动.测试与诊断,2014,34(5):802-806.
[9]程军圣,马兴伟,杨宇.基于排列熵和VPMCD的滚动轴承故障诊断方法[J].振动与冲击,2014,33(11):119-123.
[10]周涛涛,朱显明,彭伟才,等.基于CEEMD和排列熵的故障数据小波阈值降噪方法[J].振动与冲击,2015,34(23):207-211.
[11]孙伟,熊邦书,黄建萍,等.小波包降噪与LMD相结合的滚动轴承故障诊断方法[J].振动与冲击,2012,31(18):153-156.
[12]杨文志,马文生,任学平.小波包降噪方法在滑动轴承故障诊断中的应用研究[J].噪声与振动控制,2009(4):50-53.
[13]杨晨,阎树田,贺成柱,等.基于峭度与小波包络分析的滚动轴承故障诊断[J].机械制造,2014,52(2):62-64.
[14]WIDODO A,SHIM M C,CAESARENDRA W,et al.Intelligent Prognostics for Battery Health Monitoring Based on Sample Entropy[J].Expert Systems with Applications,2011,38(9):11763-11769.
[15]张晓涛,唐力伟,王平,等.基于排列熵的电磁声发射信号到达时间识别[J].现代制造工程,2015(5):126-130.
[16]郑近德,程军圣,杨宇.多尺度排列熵及其在滚动轴承故障诊断中的应用[J].中国机械工程,2013,24(19):2641-2646.
[17]冯辅周,饶国强,张丽霞,等.基于EMD和排列熵的轴承异常检测方法研究[J].轴承,2013(2):53-56.
[2]MA J,WU J D,FAN Y G,et al.The Rolling Bearing Fault Feature Extraction Based on the LMD and Envelope Demodulation[J]Mathematical Problems in Engineering,2015,2015(6):1-13.
[3]TIAN Y,WANG Z L,LU C.Self-adaptive Bearing Fault Diagnosis Based on Permutation Entropy and Manifold-based Dynamic Time Warping[J/OL].http://doi.org/10.1016/j.ymssp.2016.04.028.
[4]程军圣,史美丽,杨宇.基于LMD与神经网络的滚动轴承故障诊断方法[J].振动与冲击,2010,29(8):141-144.
[5]颜天晓,张瑞亮,王铁,等,基于EEMD和Hilbert包络分析的轴承复合故障诊断研究[J].机械传动,2016(6):132-135.
[6]杨斌,程军圣.基于WPD-LMD和排列熵的结构损伤识别方法[J].湖南大学学报(自然科学版),2014,41(8):41-46.
[7]冯辅周,饶国强,司爱威,等.排列熵算法研究及其在振动信号突变检测中的应用[J].振动工程学报,2012,25(2):221-224.[J].
[8]郑近德,程军圣,杨宇.基于LCD和排列熵的滚动轴承故障诊断[J].振动.测试与诊断,2014,34(5):802-806.
[9]程军圣,马兴伟,杨宇.基于排列熵和VPMCD的滚动轴承故障诊断方法[J].振动与冲击,2014,33(11):119-123.
[10]周涛涛,朱显明,彭伟才,等.基于CEEMD和排列熵的故障数据小波阈值降噪方法[J].振动与冲击,2015,34(23):207-211.
[11]孙伟,熊邦书,黄建萍,等.小波包降噪与LMD相结合的滚动轴承故障诊断方法[J].振动与冲击,2012,31(18):153-156.
[12]杨文志,马文生,任学平.小波包降噪方法在滑动轴承故障诊断中的应用研究[J].噪声与振动控制,2009(4):50-53.
[13]杨晨,阎树田,贺成柱,等.基于峭度与小波包络分析的滚动轴承故障诊断[J].机械制造,2014,52(2):62-64.
[14]WIDODO A,SHIM M C,CAESARENDRA W,et al.Intelligent Prognostics for Battery Health Monitoring Based on Sample Entropy[J].Expert Systems with Applications,2011,38(9):11763-11769.
[15]张晓涛,唐力伟,王平,等.基于排列熵的电磁声发射信号到达时间识别[J].现代制造工程,2015(5):126-130.
[16]郑近德,程军圣,杨宇.多尺度排列熵及其在滚动轴承故障诊断中的应用[J].中国机械工程,2013,24(19):2641-2646.
[17]冯辅周,饶国强,张丽霞,等.基于EMD和排列熵的轴承异常检测方法研究[J].轴承,2013(2):53-56.