Nonparametric probabilistic forecasting-driven electricity market clearing via improved two-stage interval optimization

The inherent variability of renewable generation poses significant operational challenges to power systems. While advanced probabilistic forecasting approaches can quantify the renewable generation uncertainty in nonparametric forms, e.g., quantiles or intervals, their integration into the market clearing process remains technically challenging. To this end, this paper proposes a novel nonparametric probabilistic forecasting-driven market clearing approach to seamlessly integrate nonparametric forecasts into decision processes. By considering both generation costs, expected real-time dispatch costs, and operational risks, an interval-based market clearing model is established to balance operating costs and risks. In this model, the uncertainty intervals are designed as endogenous decision variables, thus optimal intervals can be automatically selected to eliminate the rigidity of fixed confidence levels. To further reduce the conservatism, an improved two-stage interval optimization method is developed by substituting the mid-point of objective interval by the expectation of operating costs. The expectation is then reformulated using constrained cost variable technique and probabilistic predictions. Eventually, the interval optimization problem is transformed into a deterministic two-stage optimization problem and then solved by applying Benders decomposition. Case study analyses validate that the proposed method seamlessly incorporates nonparametric probabilistic forecasts into market clearing mechanisms and balances system economic efficiency and operational reliability.