Adjusting Post-processing Approach for Very Short-Term Solar PV Power Forecasts

Abstract

The solar energy resources are rapidly increase their shares into the generation fleet of electric power systems. Therefore, accurate solar power forecasts is a vital requirement for an economical, reliable and stable operation of these power systems. In this paper a post-processing adjusting approach is implemented to adjust and improve the intra-hourly forecasts of the PV solar power output for three stations with different climate patterns in the United States. Thus, very short-term forecasts of multiple horizons are generated by several individual models, and then these diverse forecasts are post-processed using the adjusting approach. The performance of the adjusting approach is evaluated by adopting the out-of-sample test for intra-hour forecasts of different temporal and spatial resolutions over a year. The evaluation is also conducted with a variety of performance metrics. The results indicate that the forecasts from the adjusting approach have the highest accuracy.