Double attention aircraft engine remaining life prediction based on CNN-GRU

Multi-sensors are used to monitor the health status of aero-engines, and the dual-attention aero-engine remaining life prediction method of CNN-GRU is used to improve prediction accuracy. A signal selection method of “loss of boundary mapping ability” is adopted, and a dual attention network prediction model of CNN fused with GRU is proposed to improve the prediction performance. Combining the advantages of the above prediction methods, a series of ablation experiments and comparative experiments with the latest methods were conducted on the commercial modular aviation propulsion system simulation data set. Results: The remaining life prediction error is reduced by an average of 9.15% compared with the single attention method, and is reduced by an average of 10.19% compared with the method without incorporating the attention mechanism. Conclusion: The model in this article effectively improves the prediction accuracy.