Robust fault diagnosis in uncertain linear parameter-varying systems

The purpose of this paper is to develop a new model-based approach for solving FDI (fault detection and isolation) problems in linear parameter-varying plants. The core element of the approach is that the model describing the monitored plant as well as the internal and external perturbations, results in an uncertain linear parameter time-varying model. To solve the problem, a new scheme based on the concept of linear parameter varying (LPV) filters is considered. The proposed method results in designing a LPV fault estimator. The FDI performances are formulated using the notion of quadratic H_∞ performance. It is shown that the design problem reduces to solving a system of linear matrix inequalities (LMI). Experimental results performed on the secondary circuit of a French nuclear power plant demonstrate the efficiency of the proposed method.