Decentralized Economic Dispatch via Proximal Message Passing

Abstract

A decentralized algorithm called proximal message passing (PMP) is applied to solve the AC-OPF problem for distribution networks with distributed energy resources (DERs). The second order cone relaxation of the AC-OPF is considered in the PMP algorithm which had previously been implemented only using the linearized DC power flow. In the PMP algorithm, each node shares local information regarding power and voltage (primal variables) and nodal price (dual variables) with its neighbors to minimize the local objective function at each time step. The local objective function consists of generation costs and a penalty associated with violating power flow constraints. The solution of the optimization problem provides day-ahead schedules for the economic dispatch of DERs and generators. Simulation results are presented for a modified IEEE 13 bus system and convergence of the PMP algorithm is discussed in simulations.