摘要：AbstractThe paper applies the scenario approach to stochastic model predictive control for renewable energy systems. First, the controllable and the (quasi-periodic) uncontrollable parts are decomposed. The latter is modeled by a Box-Jenkins system with appropriately chosen inputs. For the controllable part, a linear state space model is used with an affine state-feedback controller. Several numerical experiments are presented on a public lighting microgrid, e.g., about forecasting the energy balance, the effects of various controller parametrizations, reoptimization frequencies, and discarding unfavorable scenarios. The results indicate that even a low order, time-independent controller with a slow reoptimization frequency can be efficient.
关键词：Keywordsstochastic model predictive controlchance-constrained optimizationpartially controllable systemsrandomized methodsscenario approachrenewable energy system
