Observer-based fault detection and diagnosis strategy for industrial processes

Abstract

This study presents the design of a fault detection and diagnosis (FDD) scheme, composed from a bank of two types of observers, applied to linear parameter varying (LPV) systems. The first one uses a combination of reduced-order LPV observers to detect, isolate and estimate actuators faults, and the second one consists of a set of full-order LPV unknown input observers (UIO) to detect, isolate and estimate sensors faults. The observers’ design, convergence and its stability conditions are guaranteed in terms of linear matrix inequalities (LMI). Therefore, the main purpose of this work is to provide a novelty model-based observers’ technique to detect and diagnose faults upon non-linear systems. Simulation results, based on two typical chemical industrial processes, are given to illustrate and discuss the implementation and performance of such an approach.