Enhanced RBAθ method for uncertainty quantification in time varying dataset

The high velocity and variability of wind power data introduce challenges in reliably detecting ramp events and quantifying uncertainty. To address this, we propose the enhanced Ramping Behavior Analysis (RBAθ ) framework, which extends the original RBAθ method by introducing two adaptive thresholding strategies: a statistical inference-based threshold and a Random Forest–based Markov Chain Monte Carlo (RF–MCMC) threshold. These replace static thresholds with data-driven, uncertainty-aware mechanisms. Empirical evaluation on wind capacity factor datasets shows that the enhanced RBAθ -Traditional (statistical thresholding) achieves an overall performance score of 0.82, with perfect robustness (1.00) and strong balance (0.94). Compared to existing approaches, it improves consistency by 20.7% over Sliding Window Ramp Threshold (SWRT) and balance by 88.8% over Cumulative Sum (CUSUM), while enabling reliable detection of both significant and stationary ramp events. Comparative analysis further indicates that while adaptive SWRT attains the highest overall score (0.86), the enhanced RBAθ -Traditional offers greater robustness and stability, making it a more reliable solution for wind ramp event detection.