PCA Fault Isolation Using Interval Reconstruction

Fault detection and isolation (FDI) based on principal component analysis (PCA) has been widely developed. However, PCA is used for FDI without regard to model uncertainties. In this paper, the model uncertainties being represented as interval, we propose to perform multiple fault isolation by extending the reconstruction principle to interval PCA model. Variable reconstructions can be expressed as a problem of solving a system of interval linear equations. From these reconstructions, interval structured residuals are designed in order to identify the set of faulty variables. However, the number and directions of faults being a priori unknown, a multiple fault isolation strategy is proposed in order to alleviate analyzing all combinations related to simultaneous variable reconstructions. Our innovative method is illustrated on a simulation example. The interest of taking into consideration the model uncertainties on FDI will be illustrated.