An SW-ELM Based Remaining Useful Life Prognostic Approach for Aircraft Engines

Abstract

With the rapid development of prognostics and health management (PHM), the prognostic of the remaining useful life (RUL) is gradually being used for performance management and optimization. The aerospace industry is particularly in need of this, for instance, the remaining life expectancy of aircraft engines is of great significance to guarantee the safety and reliability. However, it is hard to establish the physical model of aircraft engines with the complex degradation process, which motivates the data-driven solution to RUL prediction. In this paper, a data-driven RUL prognostic approach is proposed for aircraft engines. Key performance indicators are extracted from sensor variables through principal component analysis. The summation wavelet-extreme learning machine is used to predict the KPIs’ degradation process by iterative method, and then KPIs’ degradation states are determined by subtractive-maximum entropy fuzzy clustering to calculate the RUL of engines. To validate the prediction model, aircraft engine degradation data are used for model simulation. Compared with other algorithms, the proposed method delivers superior prediction performance.