A component-level failure detection and identification algorithm based on open-loop and closed-loop state estimators

This study presents a component-level failure detection and identification (FDI) algorithm for a cascade mechanical system subsuming a plant driven by an actuator unit. The novelty of the FDI algorithm presented in this study is that it is able to discriminate failure occurring in the actuator unit, the sensor measuring the output of the actuator unit, and the plant driven by the actuator unit. The proposed FDI algorithm exploits the measurement of the actuator unit output together with its estimates generated by open-loop (OL) and closed-loop (CL) estimators to enable FDI at the component's level. In this study, the OL estimator is designed based on the system identification of the actuator unit. The CL estimator, which is guaranteed to be stable against variations in the plant, is synthesized based on the dynamics of the entire cascade system. The viability of the proposed algorithm is demonstrated using a hardware-in-the-loop simulation (HILS), which shows that it can detect and identify target failures reliably in the presence of plant uncertainties.

► This paper presents a component-level failure detection and identification (FDI) algorithm for a class of cascade systems.
► A novel analytical redundancy technique to map the pattern of the residuals to the failure of a specific system component.
► The design of open-loop and closed-loop state estimators that can be used to generate the residuals.
► Demonstration of proposed FDI algorithm using a hardware-in-the-loop simulation on a generic mechanical system driven by a hydraulic actuator.