Distributed Optimal Active Power Dispatch Under Constraints for Smart Grids

Optimal active power dispatch is a well-studied problem in power system research. The conventional solutions are usually centralized, thus are inflexible and susceptible to single-point-failure. In this paper, a fully distributed solution for optimal active power dispatch is proposed. The proposed solution can consider not only constraints of supply–demand balance and generation bounds but also the line flow constraints. To balance computational efficiency and effectiveness, DC power flow was used to check for line flow constraint violations. In this way, both the optimality and feasibility of the obtained solution can be guaranteed. The operations of projected gradient calculations and global situational awareness acquisition are fully distributed and implemented using a multi-agent system. In the multi-agent system, an agent could have two function modules for information discovery and generation optimization. The lower-level information discovery module finds the unified price and congestion prices based on the consensus algorithm. With the discovered information, the upper-level generation optimization module adjusts generation settings based on the projected gradient algorithm. Simulation studies of power systems with different scales verify the effectiveness of the proposed solution.