Multi-Agent system for self-optimizing power distribution grids

Providing, operating and maintaining the infrastructure for continuously growing and changing electric energy systems are the technological key challenges for our society in the next years and decades. Multi-Agent Systems represent a promising approach able to realize/implement the above described functionalities and services in a power distribution network. In this paper, we present an automation agent approach and the related system architecture for handling such dynamic networks. On the one side, an agent provides monitoring and diagnostics abilities required for the robust functioning of components in the distributed environment. On the other side, to coordinate energy consumption, keep nodal voltages within predefined bounds and avoid overstressing of equipment, the system also provides anticipation services. Monitoring customer demands and applying statistical methods incorporated as behaviors in the related agents, the system is able to forecast network behavior and optimize its performance. Moreover, due to its vital importance, the system offers to the human user a complete overview of the agent's activities as well as a possibility to supervise its actions through an adequate Human Machine Interface (HMI). The presented approach is currently in the development phase and will be tested in the AIT power distribution laboratory.