Probabilistic Forecasting of Temporal Trajectories of Regional Power Production - Part 2: Photovoltaic Solar

TL;DR

This paper introduces a probabilistic model for forecasting regional photovoltaic solar power production, utilizing weather forecasts and Bayesian inference to produce calibrated, physically consistent predictive distributions over multiple days.

Contribution

It develops a multivariate Gaussian model with temporal error correlation for probabilistic PV power forecasts, addressing a gap in full predictive distribution methods.

Findings

The model provides calibrated forecasts in a case study in Germany.

Incorporating temporal correlation improves forecast accuracy.

The approach yields physically consistent trajectories.

Abstract

We propose a fully probabilistic prediction model for spatially aggregated solar photovoltaic (PV) power production at an hourly time scale with lead times up to several days using weather forecasts from numerical weather prediction systems as covariates. After an appropriate logarithmic transformation of the power production, we develop a multivariate Gaussian prediction model under a Bayesian inference framework. The model incorporates the temporal error correlation yielding physically consistent forecast trajectories. Several formulations of the correlation structure are proposed and investigated. Our method is one of a few approaches that issue full predictive distributions for PV power production. In a case study of PV power production in Germany, the method gives calibrated and skillful forecasts.