[2A3] Aero engine remaining life prediction based on grey similarity multi-scale matching
S Sun¹,², H Huang¹, J Ding¹, W Lu¹,² and D Wang³
¹Harbin Institute of Technology, China
²Guangdong Provincial Key Laboratory of Intelligent Morphing Mechanisms and Adaptive Robotics, China
³Shanghai Jiao Tong University, China
In this paper, we address the challenge of predicting the remaining useful life (RUL) of aero engines, which are critical components of aircraft that operate under increasingly extreme conditions as engine performance enhances. Ensuring the safety and reliability of these engines is paramount. To this end, we introduce a novel RUL prediction methodology that leverages grey similarity multi-scale matching. This approach employs the robust capabilities of long short-term memory (LSTM) networks for processing time-series data. Specifically, an LSTM stacked autoencoder (L-SAE) is designed to extract pivotal operational features of the engine, thereby delineating its degradation trajectory. Furthermore, the grey correlation analysis is utilised to assess the similarity between these degradation trajectories, which is complemented by a multi-timescale sliding window technique for enhanced similarity matching. Subsequently, kernel density estimation is applied to gauge the uncertainty associated with the prediction outcomes. The efficacy and superiority of our proposed method are demonstrated through the validation of the experiment study. Comparative analysis reveals that our method outperforms existing techniques in key evaluation metrics, underscoring its potential applicability to large-scale datasets. This validation not only confirms the method’s effectiveness but also its advantage in predicting the RUL of aero engines with greater accuracy and reliability.
