RESGen: Renewable Energy Scenario Generation Platform

摘要

Space-time scenarios of renewable power generation are increasingly used as input to decision-making in operational problems. They may also be used in planning studies to account for the inherent uncertainty in operations. Similarly using scenarios to derive chance-constraints or robust optimization sets for corresponding optimization problems is useful in a power system context. Generating and evaluating such spacetime scenarios is difficult. While quite a number of proposals have appeared in the literature, a gap between methodological proposals and actual usage in operational and planning studies remains. Consequently, our aim here is to propose an open-source platform for space-time probabilistic forecasting of renewable energy generation (wind and solar power). This document covers both methodological and implementation aspects, to be seen as a companion document for the open-source scenario generation platform. It can generate predictive densities, trajectories and space-time interdependencies for renewable energy generation. The underlying model works as a post-processing of point forecasts. For illustration, two setups are considered: the case of day-ahead forecasts to be issued once a day, and for rolling windows with regular updates, with application to the western part of the United States, with both wind and solar power generation.