A Time-domain Decoupling Policy for Multi-stage Robust Scheduling in a Corporate Energy System with V2G

Abstract

With the recent popularity of vehicle-to-grid (V2G) technology, more plug-in electric vehicles (PEVs) are expected to be integrated into the corporate energy system (CES) as mobile batteries. Unfortunately, the non-negligible uncertainties, such as, the stochastic driving behaviors of PEVs, restrict the implementation of V2G. In this paper, uncertainties are modelled as bounded sets, and a multi-stage robust model is proposed for the CES scheduling coordinated with V2G. Different to the typical single-stage or two-stage robust model, the multi-stage robust model follows the sequential decision process in real operation, and satisfy the non-anticipativity constraint. Decision policy method is usually used to solve the multi-stage robust model. However, the decision policy is highly coupled on time-domain, and thus increase the model complexity and computational burden. In this paper, based on the V2G capacity evaluation, we present a time-domain decoupling policy which uses two intermediate variables to cover all history information about uncertainties. With the new policy, the V2G dispatching model can be equivalently decoupled to a series of sub models which are in much smaller scales and can be solved in parallel. Numerical results verify effectiveness of the proposed model. The computational efficiency is improved compared with the partitioning policy, especially for large-scale model.