Digital Twin-Driven Remaining Useful Life Prediction for Rolling Element Bearing

Traditional methods for predicting remaining useful life (RUL) ignore the correlation between physical world data and virtual data, leading to low prediction accuracy of RUL affecting normal working rolling element bearing (REB). To solve above problem, we propose a hybrid method based on digital twin (DT) long short-term memory (LSTM). The combines high simulation capabilities DT strong processing LSTM. Firstly, develop system characteristics analysis an REB. When is implemented, can obtain theoretical value RUL. Then, experimental used train LSTM model. output actual Finally, particle swarm optimization (PSO) algorithm fuses values with case study demonstrates that greater than 97.5%, which improves performance robustness Therefore, important technology REB health management (PHM). It realizes early intervention maintenance mechanical equipment ensures safety enterprises’ production.