Hybrid CNN-BiLSTM Model with Simulated Annealing Optimization for Aircraft Engine RUL Prediction

Le résumé fourni par la source

Predictive and health management is a key research direction to ensure the reliability and safety of aerospace systems, in which the remaining useful life prediction is of core significance in supporting condition-based maintenance and reducing operation and maintenance costs. In this paper, a hybrid deep learning framework integrating Convolutional Neural Network (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) network is proposed for turbofan engine RUL prediction based on the NASA C-MAPSS dataset. The CNN part is used to extract locally degraded features from the multi-sensor time series, while the BiLSTM part captures the long-term dependence of the time series in the forward and backward directions. In the data preprocessing stage, sliding window segmentation and normalization are used to improve the generalization ability of the model. Experiments are conducted on the FD004 subset of the CMAPSS dataset, which contains 249 training and 248 testing instances, each with 21 sensor signals under 3 operating conditions. After preprocessing, 14 key sensors are retained as the degradation feature inputs. Experimental results show that this method outperforms traditional machine learning methods and some deep learning baseline models in terms of prediction accuracy and robustness. The results demonstrate that the hybrid CNN-BiLSTM architecture has significant potential for datadriven prediction of complex engineering systems.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.