Fault Detection and Isolation for a Supermarket Refrigeration System – Part Two: Unknown-Input-Observer Method and Its Extension

Abstract

AbstractThe Fault Detection and Isolation (FDI) using the Unknown Input Observer (UIO) for a supermarket refrigeration system is investigated. The original system's state Tgoods (temp. of the goods) is regarded as a system unknown input in this study, so that the FDI decision is not disturbed by the system uncertainties relevant to this state dynamic and the original system disturbance Qairload (the thermal feature of the air). It has been observed that a single UIO has a very good detection capability for concerned sensor and parametric faults. However, only the parametric fault can be isolated by using a bank of UIOs. Thereby, a complete FDI approach is proposed by combining the Extended-Kalman-Filter (EKF) and UIO methods, after an extensive comparison of KF-, EKF- and UIO-based FDI methods is carried out. The simulation tests show that the complete FDI approach has a good and successful FDI capability regarding to concerned fault scenarios.