Principle and evaluation of a self-Adaptive Multi-Agent System for State Estimation of electrical distribution network

In our daily life, electricity becomes more and more important. Mainly for environmental concerns, governments tend to encourage integration of generators which rely on renewable energy sources. Therefore, it is necessary to move from present electrical networks to smarter ones. These issues have contributed to the development of the concept of Smart Grid. One might well ask the question: how to add a layer of intelligence on electrical networks as we know them? The work presented in this paper is mainly focused on the State Estimation as a way to observe the evolution of low voltage or medium voltage disturbances in order to mitigate them using innovative regulation functions. The Atena platform presented in this study has shown the feasibility and some advantages of using an adaptive multi-agent system for the estimation of the network state within a reasonable time, with encouraging accuracy for voltage regulation, with a linear complexity and with the capacity to adapt itself to changes that can occur in the network. Each agent has only a local perception of the grid and interacts with its immediate neighbors according to the network topology, without any information on the global state.