Power Supply Scheduling Optimization from a Viewpoint of Spatio-Temporal Aggregation

Abstract

With increased attention of renewable energy, a large number of photovoltaic (PV) power generators are expected to be installed into power systems in Japan. In this situation, we consider a problem to make an appropriate schedule on the following day based on the prediction of demand and PV power generation. Since the PV/demand prediction includes non-negligible uncertainty, we need to devise a method for the power supply scheduling explicitly taking into account the prediction uncertainty. Towards a robust power supply scheduling tolerating the prediction uncertainty, first, we introduce spatio-temporal aggregation and provide a fundamental fact on it. Based on this, we show that the scheduling problem can be divided into two subproblems that involve spatially and temporally aggregated variables, respectively. Then, investigating that spatio-temporal aggregation has potential to reduce the influence of the prediction uncertainty, we show that the feasibility of the scheduling problem is improved by the spatio-temporal aggregation. Finally, we show the efficiency of the power supply scheduling based on the spatio-temporal aggregation through a numerical simulation.