A Dual-Stage Attention-Based LSTM Neural Network for Tool Remaining Useful Life Prediction

As an indispensable part of the mechanical manufacturing system, the wear state of the tool will affect the productivity and machining quality. Therefore, it is of great significance to accurately predict its remaining useful life (RUL). This paper proposes a non-autoregressive dual-stage attention-based long-short term memory network (DA-LSTM) for tool RUL prediction. The dual-stage attention mechanism of the proposed method can adaptively assign attention weights to the input features and the context vectors output by the LSTM encoder, thereby improving the performance of the model without manual feature selection. Validation experiments are conducted on a real public dataset, and the experimental results show that the method proposed in this paper has better performance in short-term and medium-term RUL prediction tasks.