Impact of COVID-19 lockdown in short-term load forecasting

Accurate prediction of electrical demand is crucial for the efficient operation of power systems. However, the unprecedented activity restrictions imposed during the pandemic led to unforeseen disruptions in electrical consumption, challenging the predictive capabilities of existing systems. This phenomenon was widespread, affecting power systems globally, as evidenced by analyses of the Spanish electricity grid presented in this paper. The precision of prediction systems significantly diminished upon the implementation of activity restrictions. This article offers an in-depth analysis of the impact on prediction accuracy in the Spanish context. Moreover, it proposes a method to identify situations where the prediction system is out of control, necessitating corrective measures. The paper introduces a straightforward corrective measure that reduces errors during out-of-control periods. It is designed as an addition to the existing forecast model, modifying its output when an out-of-control situation is reached. The results suggest that further investigation could substantially mitigate the impact of such events and enhance prediction system resilience. • COVID-19 significantly impacted short-term load forecasting accuracy. • A method detects in real-time when a forecast is out of control. • No external data is needed, only the model’s own input data. • A simple correction reduces errors during disruption periods. • Tested in Spain and Germany, improving accuracy without model changes.