Distributionally Robust Energy and Reserve Dispatch With Distributed Predictions of Renewable Energy

Abstract

This article proposes a novel distributionally robust energy and reserve dispatch model with distributed renewable predictions. Through leveraging the prediction information from both the system operator and renewable energy sources, the renewable energy can be predicted more precisely, and hence the dispatch decision is improved. The proposed model captures the relationship between the expectation, variance, covariance of renewable energy and the predictive decision, leading to the formulation of a less conservative moment-based ambiguity set for renewable energy. To solve the distributionally robust dispatch with predictive decision-dependent uncertainty, we first relax the second-stage recourse value function with dual vertices, and then transform the dispatch model to a tractable form using duality theory and S-lemma. A tailored two-layer iterative algorithm is finally developed to solve the tractable model, where the outer-layer iteration solves the master and sub problems alternately to update dual vertices, while the inner-layer iteration convexifies the master problem with nonlinear constraints using alternate optimization and difference-of-convex optimization. Moreover, two acceleration strategies are developed to improve the convergence of the solution. Simulations in three testing systems validate the efficiency of the proposed model and solution algorithm.