Ensuring Cyberattack-Resilient Load Forecasting with A Robust Statistical Method

Cyberattacks in power systems can alter load forecasting models' input data. Although extreme outliers that fail to follow regular patterns can be easily identified, other more carefully-designed attacks can escape detection and seriously impact load forecasting. While existing work mainly focuses on enhancing attack detection, we propose a cyberattack-resilient load forecasting model that is based on an adaptation of classic Huber's robust statistical method. In a large-scale simulation study, the proposed method performed better than the classic method in various settings.