Data-Driven Uncertainty Sets: Robust Optimization with Temporally and Spatially Correlated Data

Robust optimization has gained increasing attention in the power system area due to its ability to model uncertainties using modest information while producing reliable solutions. However, a common concern with the robust approach is that it can be overly-conservative. To address this issue, we propose a data-driven method to construct uncertainty sets by using autoregressive integrated moving average (ARIMA) model and whitening transform on temporally and spatially correlated data. We apply the data-driven uncertain sets to the robust unit commitment problem in the presence of uncertain interchange levels. Our numerical experiments on the ISO New England system show that the proposed method outperforms traditional uncertainty sets in terms of reducing costs while maintaining system reliability.