Hyperelliptic Kalman filter-based aeroengine sensor fault FDIA system under multi-source uncertainty

Abstract

Aeroengine faces multi-source uncertainty while operating. To reduce the conservativeness and improve the performance of the diagnosis system in complex environments, the multi-source uncertainty environment, which consists of aeroengine epistemic uncertainty and stochastic uncertainty of the control system, is considered in this paper. Based on polynomial chaos expansion (PCE), the uncertainty of the state response and output response is quantified. In addition, by proposing the hyperelliptic Kalman filter (HeKF), the optimal estimation of state correlation and output variance is realized, and the adaptive weighting matrix is provided on this basis. Combined with the thresholds calculated by design indicators of the diagnosis system, conservativeness-reduced fault detection is achieved. Ultimately, based on the fault matching idea, the hyperelliptic Kalman filter bank (HeKFB) is established and implemented for fault isolation and accommodation. Open-loop and closed-loop numerical simulations demonstrate that the proposed HeKF-based aeroengine sensor fault detection, isolation, and accommodation (FDIA) system is efficient under multi-source uncertainty conditions. Furthermore, compared to the extended Kalman filter (EKF) and the Kalman particle filter (KPF), the HeKF-based FDIA system is less conservative, more sensitive to minor faults, and superior in guaranteeing the security and reliability of aeroengine operation.