Research on Prediction of Remaining Useful Life of Wheel Machining Tool Based on EMD–PSO–LSTM Model

ABSTRACT This paper proposes a hybrid EMD–PSO–LSTM model for predicting the remaining useful life of cutting tools in wheel machining. To handle the non‐stationary and noisy nature of raw sensor signals, empirical mode decomposition (EMD) is first employed as a preprocessor to decompose the signals into a set of intrinsic mode functions, thereby extracting discriminative multi‐scale temporal features. These refined features are then fed into a long short‐term memory (LSTM) network to capture the long‐term temporal dependencies inherent in the tool's degradation process. Furthermore, the particle swarm optimization (PSO) algorithm is integrated to automatically optimize the key hyperparameters of the LSTM network, enhancing model performance and convergence. Experimental results demonstrate that the proposed EMD–PSO–LSTM framework significantly outperforms benchmark models, achieving superior performance in terms of prediction accuracy, robustness, and convergence speed. This validates the model's efficacy as a reliable solution for intelligent tool condition monitoring and proactive maintenance planning in manufacturing.