A Projection-based Dual Algorithm for Fast Computation of Control in Microgrids

Abstract

We present a novel algorithm for the distributed computation of optimal predictive storage and reactive power control in microgrids. This algorithm uses a dual decomposition approach to deal with coupling constraints, and a primal projection procedure to handle local constraints. This significantly decreases the amount of iterations needed for practical convergence, as compared to the dual decomposition algorithm involving all constraints. Simulations compare the algorithm with the dual decomposition approach in four different testbeds, to show how the speed of convergence is improved.