Multimodel based fault tolerant control of the 3-tank system

The problem of fault-tolerant control of a 3-tank system benchmark process using multiple models and switching control is addressed in this paper. According to this approach, the possible modes of process operation, either due to changes in production regime and disturbances or to the occurrence of faults in sensors and/or actuators, are modelled by local linear predictive models. It is assumed that for each set of operating conditions there is a local controller which stabilizes it while embodying corresponding constraints. The fault-tolerant controller is made up by these local controllers, which commute among themselves according to the decision of a supervisor. The supervisor makes the decision by comparing the candidate local model behavior with the one of the plant and choosing the controller corresponding to the best fitted model. A minimum dwell-time for each controller is repected for stability reasons.