基于Dropout-CNN的滚动轴承故障诊断研究
|
摘要
针对滚动轴承故障特征很难提取及传统故障诊断方法准确率偏低的问题,提出一种基于Dropout的改进卷积神经网络(Dropout-CNN)结构,可以无需预先提取滚动轴承振动信号的故障特征,直接端到端的实现滚动轴承故障诊断。该方法以振动信号为监测信号,使用傅里叶变换生成振动信号频谱图,以此作为整个系统的输入,利用卷积神经网络强大的特征提取能力可以自动完成故障特征提取以及故障识别。试验结果表明该方法平均诊断准确率高达99. 5%。该方法实现了大量样本下滚动轴承不同故障类型的故障特征自适应提取与故障状态的准确识别。
Aiming at the difficulty of the rolling bearing fault feature extraction and the low accuracy of the traditional fault diagnosis method,an improved convolution neural network (Dropout-CNN) structure based on Dropout was proposed to realize the fault diagnosis of rolling bearing directly from end to end without extracting the fault features of rolling bearing vibration signals in advance. The vibration signal was used as the monitoring signal and the frequency spectrum of the vibration signal was generated by Fourier transform. The fault feature and fault identification can be automatically completed by using the powerful feature extraction capability of the convolutional neural network. The experimental results show that the average diagnostic accuracy of this method is as high as 99. 5%. It realizes adaptive fault feature extraction and fault state identification of different fault types of rolling bearings under a great quantity of samples.
Aiming at the difficulty of the rolling bearing fault feature extraction and the low accuracy of the traditional fault diagnosis method,an improved convolution neural network (Dropout-CNN) structure based on Dropout was proposed to realize the fault diagnosis of rolling bearing directly from end to end without extracting the fault features of rolling bearing vibration signals in advance. The vibration signal was used as the monitoring signal and the frequency spectrum of the vibration signal was generated by Fourier transform. The fault feature and fault identification can be automatically completed by using the powerful feature extraction capability of the convolutional neural network. The experimental results show that the average diagnostic accuracy of this method is as high as 99. 5%. It realizes adaptive fault feature extraction and fault state identification of different fault types of rolling bearings under a great quantity of samples.
引文
[1]雷亚国,何正嘉.混合智能故障诊断与预示技术的应用进展[J].振动与冲击,2011,30(9):129-135.
[2]雷亚国,贾峰,周昕,等.基于深度学习理论的机械装备大数据健康检测方法[J].机械工程学报,2015,51(21):49-56.
[3]李国杰,程学旗.大数据研究:未来科技及经济社会发展的重大战略领域———大数据的研究现状与科学思考[J].中国科学院院刊,2012,27(6):647-657.
[4] HINTON G E,SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science,2006,313(5786):504-507.
[5] LECUN Y,BENGIO Y,HINTON G. Deep learning[J]. Nature,2015,521(7553):436-444.
[6] KRIZHEVSKY A,SUTSKEVER I,HINTON G E. Image Net classification with deep convolutional neural networks[C]//International Conference on Neural Information Processing Systems.Lake Tahoe,Nevada,United States:Curran Associates Inc,2012,1:1097-1105.
[7] DU Wenliao,TAO Jianfeng,LI Yanming,et al. Wavelet leaders multifractal features based fault diagnosis of rotating mechanism[J]Mechanical Systems&Signal Processing,2014,43(1/2):57-75.
[8] LI Weihua,ZHANG Shaohui,HE Guolin. Semisupervised distancepreserving self-organizing map for machine-defect detection and classification[J]. IEEE Transactions on Instrumentation&Measurement,2013,62(5):869-879.
[9] VAN M,KANG H J. Bearing defect classification based on individual wavelet local fisher discriminant analysis with particle swarm optimization[J]. IEEE Transactions on Industrial Informatics,2017,12(1):124-135.
[10] WORDEN K,STASZEWSKI W J,HENSMAN J J. Natural computing for mechanical systems research:a tutorial overview[J].Mechanical Systems&Signal Processing,2011,25(1):4-111.
[11]曾雪琼.基于卷积神经网络的变速器故障分类识别研究[D].广州:华南理工大学,2016:9-14.
[12]李亚飞,董红斌.基于卷积神经网络的遥感图像分类研究[J].智能系统学报,2018,13(4):550-556.
[13]张伟.基于卷积神经网络的轴承故障诊断算法研究[D].哈尔滨:哈尔滨工业大学,2017:26-29.
[14] LEI Y G,ZUO M J. Gear crack level identification based on weighted K nearest neighbor classification algorithm[J]. Mechanical Systems and Signal Processing,2009,23(5):1535-1547.
[2]雷亚国,贾峰,周昕,等.基于深度学习理论的机械装备大数据健康检测方法[J].机械工程学报,2015,51(21):49-56.
[3]李国杰,程学旗.大数据研究:未来科技及经济社会发展的重大战略领域———大数据的研究现状与科学思考[J].中国科学院院刊,2012,27(6):647-657.
[4] HINTON G E,SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science,2006,313(5786):504-507.
[5] LECUN Y,BENGIO Y,HINTON G. Deep learning[J]. Nature,2015,521(7553):436-444.
[6] KRIZHEVSKY A,SUTSKEVER I,HINTON G E. Image Net classification with deep convolutional neural networks[C]//International Conference on Neural Information Processing Systems.Lake Tahoe,Nevada,United States:Curran Associates Inc,2012,1:1097-1105.
[7] DU Wenliao,TAO Jianfeng,LI Yanming,et al. Wavelet leaders multifractal features based fault diagnosis of rotating mechanism[J]Mechanical Systems&Signal Processing,2014,43(1/2):57-75.
[8] LI Weihua,ZHANG Shaohui,HE Guolin. Semisupervised distancepreserving self-organizing map for machine-defect detection and classification[J]. IEEE Transactions on Instrumentation&Measurement,2013,62(5):869-879.
[9] VAN M,KANG H J. Bearing defect classification based on individual wavelet local fisher discriminant analysis with particle swarm optimization[J]. IEEE Transactions on Industrial Informatics,2017,12(1):124-135.
[10] WORDEN K,STASZEWSKI W J,HENSMAN J J. Natural computing for mechanical systems research:a tutorial overview[J].Mechanical Systems&Signal Processing,2011,25(1):4-111.
[11]曾雪琼.基于卷积神经网络的变速器故障分类识别研究[D].广州:华南理工大学,2016:9-14.
[12]李亚飞,董红斌.基于卷积神经网络的遥感图像分类研究[J].智能系统学报,2018,13(4):550-556.
[13]张伟.基于卷积神经网络的轴承故障诊断算法研究[D].哈尔滨:哈尔滨工业大学,2017:26-29.
[14] LEI Y G,ZUO M J. Gear crack level identification based on weighted K nearest neighbor classification algorithm[J]. Mechanical Systems and Signal Processing,2009,23(5):1535-1547.