A Duality-driven Real-time Dispatch Policy for Remote Wind-storage Plant

Abstract

The utilization of energy storage systems has been popularized in renewable plant planning and dispatch. Introducing energy storage can provide system backup, smooth energy use, and reduce renewable power spillage. In this paper, we study the real-time dispatch problem of a wind-storage integrated power plant in an online optimization fashion, that is, only information observed in the current period is considered when producing real-time dispatch strategy, and predictions on the uncertain issues in the future time slots are avoided. The dispatch procedure is formulated based on partial duality of the original dispatch optimization problem. Dual variables are updated through a statistical learning procedure where only small-scale convex programs has to be solved. The proposed model can yield a feasible sub-optimal solution, and the optimality gap to the perfect information dispatch strategy is bounded. Case study corroborates the effectiveness of the proposed method. CCS CONCEPTS• Applied computing → Engineering.