Online Detection Of Incipient Fault Of Rolling Bearing Based On Deep Learning

Rolling bearing as an important part of mechanical,it is prone to multiple faults under complex working conditions such as heavy load and strong impact,which directly leads to deterioration of the working conditions of mechanical equipment.The bearing incipient fault detection techniques based on vibration signals are able to detect the health condition of rolling bearing.And further diagnosing the bearing fault and avoiding serious consequences caused by the deterioration of the mechanical equipment.Although researchers have made substantial progress in bearing fault detection and diagnosis recently,incipient fault detection,especially online detection,is still at an initial stage.Generally speaking,online detection of incipient faults is still subject to the following challenges: 1)improving discriminative ability of incipient fault features;2)adaptive recognition of the distribution inconsistency that exists in online sequential data;3)achieving automatic detections with avoiding manual adjustment of detection criterion;and 4)reducing false alarm rate.Aiming at the above four problems,This paper starts with the construction of online detection models and incipient fault characterization capabilities,and introduces semi-supervised learning and deep transfer learning,respectively.Realize the adaptive online data changes and the effective migration of incipient fault information across operating conditions.In order to solve the problems of insufficient online data incipient fault characteristics,the online detection results are late and the false alarm rate is large.The main research contents are as follows:(1)In order to adapt to the dynamic change of online data,an Online Detection of Bearing Incipient Fault with Semi-supervised Architecture and Deep Feature Representation is proposed based on the online detection model.This method has a good detection effect in the case of insufficient online data.First,we extract deep features using stacked denoising auto-encoder from the target bearing’s normal state data and an auxiliary bearing’s fault state data.Second,we introduce safe semi-supervised support vector machine(S4VM),a kind of semi-supervised classifier,to identify the sequentially arrived data of the target bearing as normal or anomalous.To update the classifier effectively,we use the principal curve to generate synthetic fault data for keeping data classes balanced during online conditionmonitoring.Finally,we propose a new fault alarm criterion based on S4 VM generalization error upper bound to adaptively recognize the occurrence of an incipient fault.The experimental results on three datasets(IEEE PHM Challenge 2012,IMS and XJTU-SY)demonstrate the effectiveness and high reliability of the proposed approach。(2)In order to enhance the discrimination ability of incipient fault features,from the perspective of transfer learning,an Online Detection of Bearing Incipient Fault based on Deep Transfer Learning is proposed.First,a new deep auto-encoder network with multi-domain transferring is proposed by constructing a new loss function with maximum mean discrepancy regularizer and Laplace regularizer.This model can extract adaptively the common feature representation among different domains,and improve effectively the feature difference between normal state and incipient fault state as well.Second,with the obtained feature representation,a new online detection model based on temporal anomaly pattern is proposed.By utilizing the permutation entropy of normal state of offline bearings to build an alarming threshold,this model can match quickly anomaly sequence in online data,and then improve the reliability of detection results.The experimental results on XJTU-SY bearings dataset show that,compared with some state-of-the-art methods of incipient fault detection,the proposed approach can obtain better real-time detection performance and lower false alarm rate.In the case of insufficient online data,this paper starts from the two perspectives of model adaptive update and incipient fault feature migration,while improving the real-time detection results and effectively reducing the false alarm rate.This provides an effective solution to the online condition monitoring and health management problems of rolling bearings,and also provides a useful reference for real-time fault warning of various rotating machinery.It has significant academic research and engineering application value.