Multi-agent based distributed power flow calculation

Abstract

Against the background of the increasing amount of distributed generation and the intention of local utilities to offer system services at the distribution level, a high degree of automation gets more and more necessary. Right now DG sources already are able to offer system services although these potentials often remain unused. Local utilities face these new challenges by applying energy management systems which essentially need information on the system state. Unfortunately, measurement is sparsely spread in distribution grids so the required data normally cannot be provided. The authors propose a new approach on managing distribution systems based on adaptive agents which are placed at different locations on the grid. They are able to locally control sources, loads and switches to keep the system state within tolerable bounds. In doing so, knowledge of the entire grid state becomes obsolete. For this purpose a decentralized power flow method is introduced. Depending on available local information each agent is able to calculate a system state. By communicating their results other agents can now react adequately. Furthermore, it is possible to infer on the entire system state as well.