摘要
负载在滚动轴承的运行过程中通常是变化的,针对变负载下滚动轴承不同故障位置及不同性能退化程度多状态识别困难的问题,提出一种基于集合经验模态分解–希尔伯特(ensemble empirical mode decomposition-Hilbert,EEMD-Hilbert)包络谱和深度信念网络(deep belief network,DBN)的滚动轴承状态识别方法。该方法首先对滚动轴承各状态振动信号进行EEMD,然后选取敏感本征模态函数(intrinsic mode function,IMF),并对其进行Hilbert变换求取包络谱。最后将各状态振动信号的IMF包络谱按顺序构建新的高维数据,输入到经遗传算法优化各隐藏层节点结构的DBN中,实现变负载下滚动轴承的多状态识别。实验结果表明:在运用DBN进行滚动轴承10种状态识别过程中,训练数据采用某种负载,测试数据选用其他负载的情况下,EEMD-Hilbert包络谱比时域或频域幅值谱能更好地体现出滚动轴承不同负载下的多状态特征;且DBN相对浅层学习的支持向量机和BP神经网络算法,具有更高的识别率,各数据集识别率均达到92.5%以上。
The load usually is changing during the running process of the rolling bearing. Aiming at the multiple-state recognition difficult problem of different fault locations and different performance degradation degrees for rolling bearing under variable load, based on ensemble empirical mode decomposition-Hilbert （EEMD-Hilbert） envelope spectrum and deep belief network （DBN）, a state recognition method of a rolling bearing is proposed. At first, the vibration signals of each state of the rolling bearing are decomposed by EEMD. Then, the sensitive intrinsic mode function （IMF） components of each signal are selected and their envelope spectrum can be obtained using Hilbert transform. At last, the new high-dimensional data can be constructed by IMF envelope spectrum of each state vibration signal according to certain order, and then the new high-dimensional data are used as the input of DBN, whose each hidden layer node structure had been optimized using genetic algorithm, and multiple-state recognition of the rolling bearing under variable load can be achieved. The experimental results show that, by using DBN, during the recognition process of the 10 states of the rolling bearing, if training data choose a certain load and testing data choose another load, the multiple-state feature of rolling bearing under different loads can be reflected better by using EEMD-Hilbert envelope spectrum than the time domain orfrequency domain amplitude spectrum. And, comparing with the shallow learning support vector machine and BP neural network algorithm, DBN has a higher recognition rate and the recognition rate of each data set can reach more than 92.5%.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2017年第23期6943-6950,共8页
Proceedings of the CSEE
基金
国家自然科学基金项目(51305109)
黑龙江省青年科学基金项目(QC2014C075)
哈尔滨理工大学青年拔尖创新人才资助项目(201511)~~
关键词
变负载
滚动轴承
集合经验模态分解
深度信念网络
状态识别
variable load
rolling bearing
empirical modedecomposition
deep belief network
state recognition
