Design of fault diagnosis filters: A multi-objective approach

In this paper, fault detection and isolation (FDI) in linear uncertain dynamic systems is addressed. The main contributions consist of the formulation of the FDI problem as a filter-based multi-objective optimization problem and the study of the tools used for the solution. The design objectives are formulated in terms of H∞, H- and generalized H2 performance specifications as well as regional constraints on filter poles. The problem is solved using linear matrix inequality (LMI) and generalized structured singular value (μg) techniques. Special design features are illustrated through a simulation example and experimental results from a controlled hydraulic process are provided to demonstrate the potential of the proposed procedure.