Distributed optimization for shipboard smartgrid

Shipboard smartgrid (SSG) requires optimized power delivery under faults. This is essentially achieved by reconfiguring the switch states (ON/OFF) in a manner that ensures maximal power to be delivered to loads after the occurrence of one or more faults. This paper seeks to investigate (1) the quality of the SSG reconfiguration solution without complete knowledge of the overall system state, and (2) Communication costs that may affect the quality of the reconfiguration. To this end, a dual decomposition based distributed optimization method for shipboard smartgrid system is proposed. The shipboard system is decomposed into multiple separable subsystems with agents. The agents locally maximize the power delivery to loads for a subsystem. Specifically, each agent solves a concave dual function of the original objective subject to convex constraints. Neighboring agent subsystems interact through convex coupling constraints and fuse the local optimizations into a globally optimal state. The reconfiguration results of the proposed implementation compared to the conventional centralized methods of reconfiguration demonstrate its effectiveness. We further investigate the convergence of the proposed approach under varying network delays and quantization noise. A theoretical lower bound on convergence due to the impact of communication network is presented.