Chair HS

Large scale systems

Large-scale systems are also a crucial class of systems, which require very different analysis and controller design techniques. Consider for instance an electricity network. Classically, we would basically view this as one system with multiple inputs and multiple outputs for which one controller has to be designed. This is no longer feasible in this day and age. We have many individual electricity producers and consumers who each make local decisions based on local information but even when everybody is willing to work together the limited information exchange makes this inherently decentralized and issues such as stability become quite hard. We have worked on the issue of achieving consensus and regulation of multi-agent systems where the systems share information to obtain a common objective but where the dynamics of the individual systems are not affected by each other. Think of a team of robots, which have to work together to achieve a common task.

Many systems that need to be controlled currently have structural constraints on the controller. A prime example is a controller, which consist of different components with limited or unreliable communications between channels. The research focuses initially on stabilization issues and linear systems. The ideal is to have a design methodology, which enables us to design a stabilizing controller, which is maximally robust against failure of one or more connections between the different components of the controller.