Sensor Fault Detection and Isolation without explicit model in bilinear systems

A method for multi-sensor Fault Detection and Isolation (FDI) is proposed for bilinear systems whose model structure is known but whose model parameters are not known. Thus, the proposed FDI method uses only available data: control signals and measured outputs and do not need an explicit model formulation. System output variables are expressed on a given time window as functions of input variables and initial unknown state. Under stability hypothesis, the influence of initial state may be neglected compared with inputs contribution. Furthermore, considering on-line outputs aggregation and orthogonal projection w.r.t. inputs, computable model-free residuals may be obtained. An academic example is provided to show the efficiency of the proposed method.