Distributed Optimal Generation Control of Shipboard Power Systems

Abstract : Traditionally, power system scheduling and control are separately implemented. To bridge the gap between these two activities, online adjustment of the optimal schedule is necessary. Because such adjustment degrades energy efficiency and dynamic response, it is desirable to integrate the two functions seamlessly. One possible solution is to optimize the control references directly. In this paper, a fully-distributed, multi-agent based control solution is presented to reduce the fuel consumption of shipboard power systems. Every generator has an associated agent that only communicates with its neighboring agents. With a properly-designed communication network, the solution can guarantee convergence, even during losses of the communication channel. This fully-distributed design can significantly improve the reliability and survivability of the system, especially during battle conditions. The improved subgradient based optimization solution can address both equality and inequality constraints and can provide performance comparable to that of centralized solutions. Simulation studies demonstrate the effectiveness of the proposed solution.