Analysis of large and coupled systems with methodes of complex networks

Dipl.-Ing. Andreas Geiger

January 2010 - December 2014

Land Hessen

Modern infrastructures as well as manufacturing and production plants are faced with a number of requirements to maintain and increase their competitiveness along with their sustainability. To reach this target, flexibility and adaptability have to be increased; self-optimization and the work in cycles have to be improved. An example for the described development is the change in the electrical energy supply network that is currently taking place. The previous clear distinction between energy producers and consumers will be transformed from a static design into a distributed system that enables a flexible operation.

To develop automation solutions which are able to tackle the described systems requirements, expanded system boundaries have to be considered along with the increase of crosslinking and overall complexity. This is combined with an even more rapid change in the systems affected by continuous adaption and changes. Therefore, during the development process of automation systems it is of interest to check properties of the system in advance. Properties of complicated systems are already analyzed in control technology by using decomposition in smaller, more manageable subsystems. However, the links between the subsystems are simplified, not completed or even not considered. But to get information about the overall system behavior it is required to consider all together the information, energy, material flow and all other couplings. A solution for analyzing large and complex systems is therefore to use abstract models to examine the structural behavior of the entire system combined with the robustness to changes. The spotlight of the examination should be on the properties depending on the crosslinking. The method should support existing automation approaches to develop useful automation systems.

One possibility for abstract description of complex systems with crosslinks is provided by the complex networks which use the graph theory. Complex Networks are already used in sociology, economics and biology to describe the links between elements and examine the associated properties. Target of the research project is to transfer these methods to the field of automation and control engineering. Thus the analysis of the entire system with respect to the system behavior, which is due to the links, will be possible.