Reactive multi-agent system applied to self-healing in Smart Grids

Abstract

This paper presents a decentralized algorithm for application in the smart grids self-healing problem, at the distribution level. The algorithm implementation is made using a reactive multi-agent system, which models the electrical grid in terms of autonomous agents which perform the algorithm operations in a distributed and parallel way. To validate this algorithm, two distribution network test models are used: a 15 bus model and a 33 bus model — standardized by IEEE. The results are obtained by means of computational simulation and shown in this paper, to each one of the network models. The results show that the proposed approach is able to recover all the nodes of the grid, within the simulation conditions. Moreover, it is seen that the multi-agent system directs the work load exactly to the failure point, preventing the involvement of the entire grid to the self-healing process.