기계학습 기반 스태킹 앙상블 모델의 풍력 발전량 예측 성능 평가
Evaluating the Performance of Machine Learning-Based Stacking Ensemble Models for Wind Power Prediction
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
The expansion of wind power generation facilities, a form of variable renewable energy (VRE), has increased the need for accurate wind power generation forecasting to mitigate instability in power grids. With advances in computational technology, machine-learning-based approaches for wind power prediction have gained significant attention, with ensemble models demonstrating notably high prediction accuracy. Among these, stacking models are particularly effective, as they enhance prediction performance by integrating the strengths of multiple individual models. However, the downstream effects of wind turbines can create differences in output between upstream and downstream units, highlighting the importance of evaluating prediction performance at both the individual turbine and wind farm levels to improve overall power system stability. Accordingly, this study assessed the power generation prediction performance of 15 individual turbines and the aggregate wind farm using four base stacking models namely, support vector machine, random forest, XGBoost, and artificial neural network and three combined stacking models.
Abstract
The expansion of wind power generation facilities, a form of variable renewable energy (VRE), has increased the need for accurate wind power generation forecasting to mitigate instability in power grids. With advances in computational technology, machine-learning-based approaches for wind power prediction have gained significant attention, with ensemble models demonstrating notably high prediction accuracy. Among these, stacking models are particularly effective, as they enhance prediction performance by integrating the strengths of multiple individual models. However, the downstream effects of wind turbines can create differences in output between upstream and downstream units, highlighting the importance of evaluating prediction performance at both the individual turbine and wind farm levels to improve overall power system stability. Accordingly, this study assessed the power generation prediction performance of 15 individual turbines and the aggregate wind farm using four base stacking models namely, support vector machine, random forest, XGBoost, and artificial neural network and three combined stacking models.