Fault-tolerant control by using lexicographic multi-objective optimization

The effect of component malfunction can be highly problematic for the normal operation of a control system particularly in the case of safety critical applications. Consequently, it is unrealistic to ignore during the control system design that faults can occur. It has been shown that Model-based Predictive Controllers (MPC) have good properties for fault-tolerance. However, an exhaustive analysis of faults in the real-time operation can lead to bad performance and even to controller infeasibility. In order to simplify the controller implementation and to overcome the mentioned difficulties, the control objectives for a plant under fault influence are prioritized according to required, degraded and unacceptable performances. In addition, a scheme for handling prioritized objectives based on the lexicographic multi-objective optimization together with a ℓ₁-norm-MPC is proposed.