Online selection of H∞ controllers for a faulty linear system

This paper investigates optimal selections of controllers online for a linear system with actuator faults in the framework of H_∞ control. Actuator faults are modeled into three states: normal, loss of effectiveness and outage, and the development of the actuator faults is predicted by using a Markov chain model. We present five methods to construct a smaller bank of controllers. Four selection schemes are proposed to select an optimal controller from a bank of controllers for different demands. Another selection scheme and one pre-selection scheme are presented when there may be false fault detection and isolation (FDI). Considered together, they provide a powerful solution for all kinds of demands. Finally, we utilize an example to explain and verify the proposed methods.