Bearing Fault Diagnosis Based On Vibration Signal

Rolling bearing is one of the most widely used parts in modern mechanical equipment.As the core of rotating machinery,once failure,it poses a serious threat to the normal operation of equipment.Usually,the vibration signals collected by sensors are the superposition of vibration signals of multiple components.At the same time,there are some problems in fault detection,such as insufficient sample size,complex mode and difficult to identify.Therefore,a new rolling bearing fault identification algorithm based on peak density classification is proposed to improve the accuracy of fault identification.In order to solve the problem of pattern recognition of rolling bearings,a scheme is designed,which mainly includes the following five steps.1,in the actual industrial production environment,the vibration signals of rolling bearings collected by sensors are the superposition of vibration signals of multiple components,so they will contain a lot of noise.Because of the influence of these noises and irrelevant frequencies,the experimental results will be greatly affected,so the primary task of bearing fault diagnosis is to denoise the original signal.2,in order to solve this problem,a negative entropy based blind source separation algorithm is firstly used for noise reduction.That is to say,the source signal is separated from the aliasing signal by blind source separation technology,and the optimal estimation of the source signal is obtained.3,in order to extract the key and effective bearing fault information more accurately and find the fault data with the most obvious difference from the normal vibration signal,a TARZAN abnormal sequence detection algorithm is used to detect the abnormal length of the sequence.4,through time domain,frequency domain and time-frequency analysis,several characteristic parameters representing the running status of rolling bearings are calculated.After that,principal component analysis method is used to reduce dimension to eliminate the correlation between characteristic parameters and form independent principal components to form fault feature vectors of bearings,which can accurately reflect the operation status of bearings.5,in order to solve the problem that the mode is complex and the fault situation is difficult to identify,a new classification algorithm based on peak density is proposed to identify the fault data.In order to verify the feasibility of the proposed scheme,the bearing fault data set of xichuan University and the data set collected from QPZ-11 mechanical vibration analysis and fault diagnosis platform are compared with other classical classification algorithms such as KNN,J48 and NB on different fault ratios(different ratios of fault data and normal data).The experimental results show that the proposed density peak classification algorithm can recognize more than 90%of the fault types,and to some extent,the feasibility of the proposed scheme and the accuracy of the algorithm are proved.