Prediction of Positive/Negative Error of a Day-Ahead Forecast of Aggregated PV Power Output

Abstract

To enhance the profitability of electricity retailers, minimizing the penalty cost associated with imbalances between day-ahead electricity supply plans and the actual supply on the designated day is crucial. One significant contributor to these imbalances is the forecast error related to Photovoltaic (PV) power output, especially when a considerable amount of PV is incorporated into the supply plan. A Battery Energy Storage System (BESS) can help mitigate these imbalances, although its limitations concerning available energy capacity can be more critical than those concerning available power capacity. To use the limited energy capacity of a BESS efficiently, it should be fully charged for expected positive forecast errors and fully discharged for negative ones. This study introduces a streamlined approach to predicting the sign of positive or negative (P/N) errors—specifically, whether the day-ahead forecast error of aggregated PV power output will be positive or negative. The method bases its predictions on the direction of change in forecasted insolation for subsequent initial time intervals. Our evaluation, conducted over a year using aggregated PV power data, reveals that the hit ratio for predicting the P/N error sign exceeds 70%. Furthermore, this prediction demonstrates increased reliability on days characterized by substantial absolute forecast errors. © 2026 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC.