Application of Fault Detection, Diagnosis and Fault Tolerant Control Algorithm to a Benchmark Multi Input Multi Output Quadruple Tank Process

摘要

Plethora of work has been done on different fault detection and isolation techniques but stillnot much work has been done on how to integrate such fault diagnosis technique with existing controllerfor efficient plant operation. There is a need to investigate an effective fault tolerant control system whichwill allow normal operation of the plant without causing any loss/damage until scheduled maintenance iscarried out for safety and economic reasons. In this work, we have used an active approach to faulttolerant control using a model based technique, known as generalized likelihood ratio test and a dynamicmatrix control algorithm. However, the active fault tolerant control approach which uses generalizedlikelihood ratio test, is first proposed by Willsky et. al. (1976) and later adapted by Narasimhan et. al.(2002) for identifying sensor biases and failures and actuator biases and failures. Our aim was to explorethe applicability of fault tolerant control approach to Nuclear Steam Generator model developed byAstrom et. al. (2000). The nuclear steam generator model is an open loop unstable system and exhibitsNon-Minimum Phase behavior. In this work, we have chosen to work with a benchmark multi input multioutput (MIMO) quadruple tank system proposed by Johanssan et. al. (2000) which also exhibits Non-Minimum Phase behavior. The result of simulation study shows the effectiveness and the improvedperformance of the proposed approach. However, it is assumed that the faults occur sequentially in time.