Multi-Agent based management of distribution networks

Abstract

Now, the large-scale integration of renewable energy resources and distribution generations leads to more complex power systems. To deal with this integration, new technologies based on power electronics and information and communication technologies (ICTs) have been explored to manage the new power systems, called active distribution systems. To facilitate the integration of distributed generation, active distribution networks have emerged. It is always important, at the active distribution networks management center level, to acquire accurate real time measurements to be able to control and take the appropriate decision. State estimation is an important function in distribution systems in general and active distribution networks in particular. State estimation must be done to ensure the good functioning of the active distribution system. Several authors have proposed some methods to solve the state estimation problem based on the power flow equations. Some methods are numerical like newton method or linear programming and some others based on artificial intelligence. In this thesis, we will propose a new multi-agent system-based approach. This technique is based on multi agent systems, to split the active distribution network and to manage the resulting sub-networks, and the metaheuristic algorithm ABC to perform the state calculations. Our approach is tested on IEEE 6-bus, 14-bus and 30-bus. The results show a dramatic decrease in the computational burden, thus a faster estimation in large systems can be obtained. This demonstrates the effectiveness of the proposed strategy