Fault detection and isolation methodology using interval predictors with application to DC-DC Buck converters

In this paper, the problem of Fault Detection and Isolation (FDI) in DC-DC Buck converter is addressed. In a bounded error context and taking into account the parameters uncertainties of a basic Buck converter model, a set-membership FDI methodology is proposed. A Linear Parameter Varying (LPV) form of the basic DC-DC Buck converter is firstly given. Under noisy environment and based on the proposed LPV model, an interval predictor developed in recent work for LPV systems is applied in this paper in order to detect and isolate multiple faults. An original signal is proposed to ensure the purpose with no need to a bank of residuals in multi-fault cases. The efficiency of the proposed methodology is illustrated through simulation results.