Accurately predicting the Remaining Useful Life (RUL) of aero-engines is of paramount importance for ensuring the safety and reliability of aircraft operations. Existing deep learning–based approaches primarily focus on extracting temporal features from engine data, yet they often overlook the heterogeneous importance of internal features, thereby constraining predictive accuracy. To address this limitation, this study proposes an end-to-end predictive framework that integrates Temporal Convolutional Networks (TCN) with attention mechanisms. Specifically, the raw engine data are first encoded through a self-attention mechanism; subsequently, the TCN is employed to capture high-dimensional temporal representations; finally, a Coordinate Attention (CA) mechanism is introduced to refine feature modeling along both temporal and spatial dimensions, enabling precise localization of critical information and thereby enhancing RUL prediction performance. Experimental results on the publicly available CMAPSS dataset demonstrate that the proposed method substantially outperforms state-of-the-art approaches in predictive accuracy, thereby validating its effectiveness and superiority.

Remaining Useful Life Prediction,Aero-Engines,Attention Mechanism