An off-line optimization approach for online energy storage managementin microgrid system

This paper investigates the real-time energy management in power system with distributed microgrids, which are independently operated and each is modeled to comprise of a renewable generation system, an energy storage system and an aggregated load. We jointly optimize the energy charged/discharged to/from the storage system and that drawn from the main grid over a finite horizon to minimize the total energy cost of conventional generation subject to given load and storage constraints. We assume that the renewable energy offset by the load over time, named net energy profile, is predictable but with finite errors. First, we consider the “off-line” optimization under an idealized assumption that the net energy profile is known ahead of time, and derive its optimal closed-form solution. Next, by applying the off-line solution combined with a sliding-window based sequential optimization, we propose a new “online” algorithm for real-time energy management under the practical setup with noisy predicted net energy profile subject to arbitrary errors. Finally, through simulations, we compare the performance of our proposed online algorithm against the conventional dynamic programming based solution as well as a heuristically designed myopic algorithm under a practical setup.