Three-stage hyperelliptic Kalman filter for health and performance monitoring of aeroengine under multi-source uncertainty

Abstract

Aeroengine operation is inevitably subject to multi-source uncertainty, which consists of epistemic uncertainty related to the aeroengine and stochastic uncertainty associated with the control system. This paper presents a solution for health and performance monitoring under multi-source uncertainty to ensure the safety and reliability of aeroengine whole-life operation in complex environments. Based on the hyperelliptic Kalman filter (HeKF), optimal health monitoring is achieved by treating health parameters as the augmented state. Meanwhile, the conservativeness-reduced output prediction is realized with the extra estimation of the unknown state function bias caused by probabilistic system parameters. Considering the computational effort can be significantly reduced by designing a multi-stage filter, the three-stage hyperelliptic Kalman filter (ThSHeKF) is finally developed, achieving high accuracy health parameter estimation and adaptive performance prediction under multi-source uncertainty. Open-loop and closed-loop numerical simulations demonstrate the effectiveness of the proposed ThSHeKF-based health and performance monitoring with high estimation accuracy. Furthermore, compared to the most commonly used extended Kalman filter (EKF), Monte Carlo (MC) experiments shows that the proposed ThSHeKF is less conservative, has better robustness, and is superior in adaptive monitoring under multi-source uncertainty.