A Novel RUL Prediction Method for Bearing Using IPVMD-LSTM

Abstract

With the rapid development of high-end Computer Numerical Control (CNC) machine tools, aeroengines and other large-scale mechanical equipment towards high precision and intelligence, it is an extremely important task to carry out health management of equipment and ensure the equipment can work in safety and stability. The essential part of mechanical equipment are bearings, whose performance will directly determine the health of the equipment. Predicting the remaining life of bearings can provide effective decision support for equipment maintenance plans, so as to avoid safety accidents, which is significant for the health management of mechanical equipment. Currently, signal processing methods and data-driven methods are widely used in bearing life prediction. However, mechanical equipment has been in the background of strong noise for a long time, and its feature signal extraction is difficult, and the traditional regression prediction accuracy is low. Aiming at the above problems, a bearing residual life method based on Improved Parameter Adaptive Variational Mode Decomposition-Long Short Term Memory Networks (IPVMD-LSTM) model is proposed. IPVMD-LSTM has two characteristics: (1) Fully considering the characteristics of bearing cyclostationarity and impulsiveness, a synthetic index is constructed and used as the objective function, the parameters of VMD are optimized by Particle Swarm Optimization (PSO), so as to reduce noise effect influence. (2) Fully consider the temporal characteristics of the actual working condition data, and use the LSTM to extract the temporal characteristics for prediction. The experimental results show that the IPVMD-LSTM method in this paper has a significant improvement in the prediction accuracy, and its Root Mean Square Error (RMSE) is reduced by 2.81% compared with the traditional method.