Advanced Method and Cost-based Indices for Probabilistic Forecasting the Generation of Renewable Power

Abstract

The ability to forecast photovoltaic (PV) power-production accurately and reliably is of primary importance for the appropriate management of the future distribution systems and for making decisions to satisfy the needs of all the stakeholders of the electricity energy market. Several forecasting methods have been proposed in the relevant literature and many indices have been used to quantify the accuracy of the forecasts. The majority of methods provides deterministic forecasts even though a great interest was recently dealt with probabilistic forecast methods. Similarly, the majority of indices that have been used to quantify the forecasting accuracy refers to deterministic forecasting and does not directly account for the economic consequences of forecasting errors in the framework of competitive electricity markets. In this paper, advanced, more accurate probabilistic indices are proposed: they account directly for the economic consequences of forecasting errors and the uncertainties that characterize the PV power-production. The improved capability of the proposed indices was verified on the PV power-production forecasted by using an advanced probabilistic forecasting method based on a Bayesian Inference approach. Numerical applications, that considered an actual PV plant, also are presented to provide evidence of the forecasting performances of both Bayesian-based approach and probabilistic indices that were considered.