Duality of Analytical Redundancy and Statistical Approach in Fault Diagnosis

Abstract

Two well-known methods of fault diagnosis (detection and isolation) are discussed in the paper : geometrical based (via analytical redundancy) and statistical based approaches. Both of the above approaches are used to solve the same practical problem, namely, to design a fault detection/isolation algorithm which is robust with respect to a model uncertainty, unknown states, and unmeasured disturbances and while sensitive with respect to a fault, in order to insure good detection/isolation properties. The goal of the paper is to discuss a duality of these different techniques.