Abstract The D cable, as a key on-board equipment for high-speed railway trains, accurately predicting its remaining useful life is of great significance for ensuring the safe and efficient operation of the trains. This paper proposes a RUL prediction method for the D cable based on SA-LSTM. Firstly, this paper analyzes the internal structure, main failure modes and failure mechanisms of the D cable, and builds a finite element model of the D cable based on ANSYS. Secondly, based on the thermal breakdown failure mechanism, a full life cycle accelerated degradation dataset of the D cable under different thermal stress conditions is constructed. Then, through the Random Forest (RF) algorithm, key degradation features are actively screened, and the Self Attention (SA) mechanism is introduced to integrate statistical features and combine the Long Short-Term Memory (LSTM) network to achieve accurate RUL prediction. Finally, the effectiveness of the proposed method is verified by the D cable degradation simulation dataset. Experimental results show that the proposed SA-LSTM model significantly improves prediction accuracy. Compared with traditional methods such as LSTM, GRU, SVR, and FNN, the proposed model achieves an average reduction of 52.4% in Mean Absolute Error (MAE) and 49.7% in Root Mean Square Error (RMSE) across six typical thermal stress conditions, proving its accuracy and effectiveness.
Chai et al. (Mon,) studied this question.