FAULT TOLERANT MPC WITH AN EMBEDDED FDI SYSTEM

Abstract

The fault detection and isolation (FDI) in industrial processes has been under an active study during the last decade, but fault tolerant control applications relying to the traditional FDI methods have been scarce. In this paper a fault tolerant model predictive controller (MPC) with an embedded FDI system is developed for controlling a simulated heavy oil fractionator process, the Shell Control Problem (SCP). Two different kinds of FDI systems are used for achieving the fault tolerance in co-operation with MPC: system based on Principal Component Analysis (PCA) and Partial Least Squares (PLS) and a system based on Subspace Model Identification (SMI). The effectiveness of selected methods was successfully tested by introducing bias and drift faults to simulated process measurements. Finally the results are presented and discussed.