A scenario-based wind power forecasting using multivariate probabilistic power curves for power grid congestion risk management under extreme grid conditions

Accurate wind power forecasting is essential for grid stability and congestion management in renewable-integrated power systems. Traditional deterministic methods fail to capture meteorological uncertainties, limiting their effectiveness in identifying extreme grid stress scenarios. This study proposes a multivariate Probabilistic Power Curve (PPC) framework that integrates wind speed, temperature, and humidity using Kernel Density Estimation (KDE) and Monte Carlo Simulation (MCS). The resulting percentile-based wind power scenarios represent multivariate meteorological uncertainty, allowing scenario-driven forecasting that captures both central trends and tail events. Case studies using real-world data from a wind farm in Jeju Island demonstrate that the proposed framework reduces forecasting error (up to 36.89 % NMAE) while identifying congestion risks overlooked by deterministic models. Under Extreme Grid Conditions (EGC), low-wind, high-demand scenarios show congestion probabilities exceeding 60 %, highlighting the model's ability to simulate high-impact, low-probability (HILP) events. By integrating probabilistic forecasting with grid-level congestion analysis, the proposed framework supports adaptive dispatch, risk-informed planning, and market-based strategies such as congestion pricing. This work offers a structured decision-support framework for uncertainty-aware transmission operation and planning under uncertainty. Future extensions may include real-time grid operation optimization, multi-energy system coordination, and resilience enhancement under climate-induced extreme scenarios.