Incipient Fault Detection And Condition Monitoring Of Rolling Bearings By Using Mahalanobis-Taguchi System

Rolling bearing is an important part in rotating machinery. It will lead to the equipment unplanned downtime due to its performance degradation or failure. In traditional, regular maintenance often results in "insufficient maintenance" or "excessive maintenance" due to the large dispersion of rolling bearing life. So it is significant meaning to investigate the incipient fault detection and condition monitoring of rolling bearings for long-term security, stable operation and predictive maintenance. The evolution tendency of fault and characteristic parameters of rolling bearing are analyzed in this paper. Two groups of feature parameters are fused into single feature parameters in Mahalanobis-Taguchi System(MTS). A new method of incipient fault detection and performance degradation is presented in this paper. The main contents are as follows:(1) A new method of incipient fault detection and condition monitoring is presented by MTS. The relationship between sensitivity and degradation condition of incipient fault of rolling bearing vibration signal feature parameters is analyzed. The advantages of MTS in multi-parameters fusion and abnormal samples detection is studied. The feature parameters which are sensitive to incipient fault and related performance degradation condition are regarded as a reference space for MTS. Two groups of feature parameters are fused into single feature parameters in MTS of Mahalanobis distance. The incipient fault detection and performance degradation is estimated by variation tendency of Mahalanobis distance.(2) According to the variation tendency of MD1 to detect the initial fault of bearing. The sensitive parameters of the incipi ent fault, such as kurtosis, crest factor, clearance factor, shape factor, impulse factor characteristic are selected to build a reference space for MTS. The fused MD1 is more sensitive and stability to the incipient fault of bearing. when incipient fault occcur, the value of MD1 increases rapidly and even beyond the maximum. Thus according to the variation tendency of MD1, the initial fault of rolling bearing can be detected accurately.(3) According to the variation tendency of MD2, bearing ’s running condition is monitori. The feature parameters, such as average, RMS, peak, amplitude of characteristic frequency of bearing inner ring, outer ring, rolling element and cage, which are related performance degradation condition are regarded as a reference space for MTS. The fused MD2 is related to the bearing degradation condition. It will change variously with the development of fault. Thus according to the variation tendency of MD2, the bearing’s running condition can be supervised.