Distributed solution for the economic dispatch problem

A distributed approach for the economic dispatch of generating units is presented. It is assumed that the generators are connected by a communication graph. Each unit has information about itself, and can exchange information only with a few neighboring units in such a graph. Using graph theory and consensus algorithms, the proposed approach solves the economic dispatch problem in a distributed manner instead of existing centralized approaches. The proposed solution is shown to be optimal, independent of the initial power distribution, and to respond automatically to real-time load demand changes. Moreover, it requires only that the communication graph be connected. The effectiveness of the proposed algorithm is verified by simulating a standard IEEE test system.