Fault Detection and Isolation for Non-Linear Systems Using Autonomous Multiple Models

Abstract

Real-time fault detection and isolation capability has become a competitive differential for modern, complex systems due to the increasing demand for higher levels of both reliability and safety. The use of health monitoring algorithms can be considered a powerful decision tool and a key enabler for new maintenance and logistic strategies in which decisions are made based on the estimated health condition of the systems under consideration. This paper presents the study of a fault detection and isolation (FDI) algorithm for non-linear systems based on a multiple-model architecture. The Extended Kalman Filter (EKF) is used as residual generation tool while the Autonomous Multiple Models (AMM) algorithm is used for residual evaluation. Numerical experiments are conducted to assess the performance and the limitations of the algorithm. The study covers an assessment of the algorithm sensitivity to different failure intensities and also its response to failures not initially considered in the fault detection and isolation architecture.