A Digital Twin Model of Life-Cycle Rolling Bearing With Multiscale Fault Evolution Combined With Different Scale Local Fault Extension Mechanism

The digital twin of life-cycle rolling bearing is significant for its degradation performance analysis and condition prediction. To solve the problem which is not reliable to arrange the production cycle by predicting diagnostic results in existing studies, because it is not accurate to only consider single-scale fault in the life-cycle bearing modeling. It is studied that the multiscale fault evolution law close to the true fault involves microscopic cracks, mesoscopic spall, and macroscopic defect, by establishing the life-cycle digital twin model with the outer ring fault. Based on the measured signals and the dynamic model with the outer ring fault, the time-varying 2-D sizes of multiscale faults are estimated. The dynamic mapping relationship between the fault dimensions and the measured signals is established using the BP network, and the fault progressive mechanism of the bearing in the whole life is analyzed. Then, by substituting the dynamic excitation of evolutionary fault into the mechanism model, the digital twin model of the life-cycle rolling bearing with multiscale fault is established in virtual space. The real-time update of the digital twin model is realized by integrating the real-time sensor data of faulty bearings and mapping the model subspace. The accuracy of the model is verified by comparing the digital twinning results in the time domain with the measured signals. It is reliable for the proposed model to improve the production efficiency by predicting the fault extension condition of the life-cycle rolling bearing accurately.