 |
Center for Nonlinear Studies |
Smart grid is an intelligent way of providing electricity to the customer with minimizing the losses of energy. We introduced a toy-model of the power-grid and explore the advantages of connecting customers to more than one central. In our model the goal is to find a configuration of switches to minimize the losses without overloading any of the power generators. We designed a stochastic local search algorithm called WalkGrid, we also study the performance of the so-called Belief Propagation algorithm and the related reinforcement technique. I will in a first part present the principle of the message-passing algorithm and how to apply to graph models, then present the stochastic algorithm and finally I will present the preliminary results of my work here.
Host: CNLS