A Workflow for Synthetic Data Generation and Predictive Maintenance for Vibration Data

Digital twins, virtual representations of real-life physical objects or processes, are becoming widely used in many different industrial sectors. One the main uses digital twins is predictive maintenance, and these technologies being adapted to various new applications datatypes processes. The aim this study was propose a methodology generate synthetic vibration data using twin model maintenance workflow, consisting preprocessing, feature engineering, classification training, classify faulty healthy for state estimation. To assess success proposed mentioned steps were applied publicly available dataset from twin, five state-of-the-art algorithms. For several algorithms, accuracy result achieved on public reached approximately 86%, data, 98%. These results showed great potential methodology, future work area.