A Scenario State Representation for Scheduling Deferrable Loads under Wind Uncertainty

Abstract

Integration of renewable energy in power systems is a potential source of uncertainty, because renewable generation is variable and may depend on changing and highly uncertain weather conditions. In this paper we present and evaluate a new method to schedule power-demanding tasks with release times and deadlines under uncertainty, in order to balance demand and uncertain supply. The problem is considered as a multi-agent sequential decision making problem where agents have to deal with uncertainty. Our main contribution is a scenario state representation and an algorithm that computes a belief over future scenarios, rather than states. The algorithm is used to recompute the belief when new information becomes available. Experiments show that our method
matches demand and uncertain supply to reduce grid power consumption, and outperforms an existing online consensus scheduling algorithm.