머신러닝 기반 전력 수요 예측: 모델 개발 및 핵심 특성 분석
Machine Learning-Based Power Demand Forecasting: Model Development and Key Feature Analysis
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This study aims to develop a high-performance model for forecasting UK electricity demand (TSD) from 2009 to 2024 using machine learning. We implemented Random Forest, XGBoost, and LightGBM as individual models and a Stacking Ensemble model combining them. After the hyperparameter optimization using GridSearchCV, the Stacking Ensemble model achieved the best performance with an R2 score of 0.9926. Furthermore, Explainable AI (XAI) analysis using SHAP identified 'settlement_period' (time of day), 'month', and 'year' as the most critical features. This research demonstrates the superiority of the ensemble approach and provides transparent, interpretable grounds for predictions, contributing to data-driven decision-making in power system operations.
Abstract
This study aims to develop a high-performance model for forecasting UK electricity demand (TSD) from 2009 to 2024 using machine learning. We implemented Random Forest, XGBoost, and LightGBM as individual models and a Stacking Ensemble model combining them. After the hyperparameter optimization using GridSearchCV, the Stacking Ensemble model achieved the best performance with an R2 score of 0.9926. Furthermore, Explainable AI (XAI) analysis using SHAP identified 'settlement_period' (time of day), 'month', and 'year' as the most critical features. This research demonstrates the superiority of the ensemble approach and provides transparent, interpretable grounds for predictions, contributing to data-driven decision-making in power system operations.