Dynamic output feedback control for continuous-time Markov jump linear systems with hidden Markov models

In this work, we study the design of dynamic output feedback controllers for continuous-time Markov jump linear systems in the worst-case scenario of partial observation of both, the Markov chain and the state, through an iterative method. We assume that the controller has only access to the output of a fault-detection and isolation device and present design conditions for the H-2, 'H-infinity, and mixed H-2/'H-infinity controller that switches according only to the detector. As a by-product of our formulation, we are also able to design mode-dependent and independent controllers, and briefly discuss the design of filters depending on the detector. The results are given in terms of bilinear matrix inequalities that allow for an iterative separation procedure method in which in each step, we solve a linear matrix inequality problem. We also present an illustrative example in the context of systems subject to faults.

Automated summary

In this work, the design of dynamic output feedback controllers for continuous-time Markov jump linear systems is studied. The worst-case scenario of partial observation is considered, where the controller only has access to the output of a fault-detection and isolation device. Design conditions for different controllers are presented based on the switching of the detector. Mode-dependent and independent controllers are also designed, and the design of filters depending on the detector is briefly discussed. The results are given in terms of bilinear matrix inequalities, allowing for an iterative separation procedure method.