Scholarlink

This study presents a hybrid approach for forecasting cooling and heating loads in commercial buildings by integrating the Barnacles Mating Optimizer (BMO) metaheuristic algorithm with the Extreme Gradient Boosting (XGBoost) machine learning technique. The proposed approach aims to leverage the complementary strengths of BMO’s evolutionary optimization capabilities and XGBoost’s predictive modeling prowess to enhance the accuracy and efficiency of load forecasting. The dataset utilized in this study is sourced from the UTA dataset, originating from the main campus of the University of Texas in Austin. It comprises hourly records of cooling, heating, and electricity load data, alongside meteorological factors such as dry bulb temperature, wet bulb temperature, and relative humidity. Performance evaluation is conducted using standard metrics including Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Maximum Error (MAX), and Standard Deviation (STD DEV). The results demonstrate that the hybrid BMO-XGBoost approach outperforms the selected algorithm, viz. Ant Colony Optimization (ACO-XGBoost), achieving lower error rates and greater predictive accuracy. This research contributes to advancing the field of building energy management by providing an effective solution for optimizing energy usage and maintaining occupant comfort in commercial buildings.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hybrid Barnacles Mating Optimizer-Extreme Gradient Boosting for Forecasting Cooling and Heating Loads in Commercial Buildings

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zuriani Mustaffa,
  • Mohd Herwan Sulaiman,
  • Fathihah Nabilah Mohd Azlan,
  • Mohd Mawardi Saari

摘要

This study presents a hybrid approach for forecasting cooling and heating loads in commercial buildings by integrating the Barnacles Mating Optimizer (BMO) metaheuristic algorithm with the Extreme Gradient Boosting (XGBoost) machine learning technique. The proposed approach aims to leverage the complementary strengths of BMO’s evolutionary optimization capabilities and XGBoost’s predictive modeling prowess to enhance the accuracy and efficiency of load forecasting. The dataset utilized in this study is sourced from the UTA dataset, originating from the main campus of the University of Texas in Austin. It comprises hourly records of cooling, heating, and electricity load data, alongside meteorological factors such as dry bulb temperature, wet bulb temperature, and relative humidity. Performance evaluation is conducted using standard metrics including Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Maximum Error (MAX), and Standard Deviation (STD DEV). The results demonstrate that the hybrid BMO-XGBoost approach outperforms the selected algorithm, viz. Ant Colony Optimization (ACO-XGBoost), achieving lower error rates and greater predictive accuracy. This research contributes to advancing the field of building energy management by providing an effective solution for optimizing energy usage and maintaining occupant comfort in commercial buildings.