Prediction of Remaining Useful Life of Bearing Based on Long Short-Term Memory Network Optimized by Slime Mould Algorithm

Abstract

For resolve the trouble of low prediction accuracy and poor generalization ability of the neural network in the prediction of remaining useful life (RUL) of bearing, a prediction manner because of the combination of slime mould algorithm (SMA) and long short-term memory (LSTM) was proposed. The hyperparameters of the LSTM were automatically optimized by using the SMA with dynamic search ability. The trained SMA-LSTM model was adopted to calculate the RUL of bearing. The simulation effects indicate that the optimization effect of SMA is improving than the grey wolf optimizer and genetic algorithm. The SMA-LSTM model is improving than the SMA-SVR and SMA-BP model in bearing RUL prediction, which proves the capability of the put forward method.