报告摘要：
Microgrids represent a promising paradigm forfuture electric power systems that integrates local renewable energy (e.g.,wind and solar), energy storage and possible co-generation (i.e.,combined heat and power generation) to reduce the operating cost and loadburden from the grid. There arises a subtle optimization problem of how tobalance these sources of energy. Also, there is considerable uncertaintyinherent in various inputs (e.g., renewable energy, demands, andelectricity prices). In this talk, we will investigate the cost minimizationproblem by modeling it as a stochastic mixed-integer programming problem, whichis challenging to solve. We devise an effective online algorithm that does notrely on the a-priori knowledge of stochastic distribution of the inputs, and isbased on a two-timescale Lyapunov optimization approach. We show that theproposed algorithm is efficient with very low computational complexity and isable to achieve near-optimal performance. Moreover, extensive empiricalevaluations using real-world data are provided to study the effectiveness ofthe proposed algorithm in practice.
