Smart Digital Twin-Based Bearing Fault Pattern Recognition

Abstract

In this research, the combination of the smart digital twin (SDT) and the machine learning technique is prescribed to have a reliable fault pattern recognition in this effort. In the first stage, the SDT for the bearing is designed by the dynamical system modeling, updated using the data-driven autoregression approach, and estimate the performance using smart Kalman filter (SKF). Thus, first, the data-driven-based autoregressive is selected to update the mathematical model of bearing and design an effective modeling section of the digital twin. Next, the SKF for the bearing signal estimation is designed by the combination of the Kalman Filter and fuzzy logic approach. In the second stage, the difference between original and estimated signals are computed. Finally, in the last stage, the support vector clustering (SVC) is recommended for clustering the bearing’s situations. The precision of the proposed procedure for the bearing fault pattern recognition is around 97.8%.