Scholarlink

<p>Heating energy is a significant component of building energy consumption and is mostly derived from fossil fuels. Predicting the amount of energy used for building heating is a critical first step in improving the energy consumption situation, particularly in developed and emerging nations, given the construction sector’s significant contribution to energy consumption. In order to anticipate heating energy consumption in various building types on an hourly basis, this work aims to develop a hybrid model that makes use of hybridization capabilities. XGBoost, HGBoost, CatBoost, and LightGBoost are four boosting algorithms that were combined with the GWO algorithm to create four different hybrid models. Optimizing and modifying the boosting algorithms’ hyperparameters was the aim of this hybridization. The case study’s findings demonstrated the appropriate accuracy of the suggested methodology for hourly heating energy consumption forecast. The XGBoost-GWO hybrid model has the best evaluation index values in commercial and residential structures, whereas the CatBoost-GWO hybrid model has the best evaluation index values in big venue buildings, according to the research findings. Thus, these models are recommended for predicting heating energy usage on an hourly basis.</p>

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

A hybrid model approach for accurate hourly forecasting of heating energy consumption for different types of urban buildings

  • 期刊论文

  • 发表日期：

    2025年9月22日

查看全文
  • Wenguo Chen,
  • Mengmeng Xu

摘要

Heating energy is a significant component of building energy consumption and is mostly derived from fossil fuels. Predicting the amount of energy used for building heating is a critical first step in improving the energy consumption situation, particularly in developed and emerging nations, given the construction sector’s significant contribution to energy consumption. In order to anticipate heating energy consumption in various building types on an hourly basis, this work aims to develop a hybrid model that makes use of hybridization capabilities. XGBoost, HGBoost, CatBoost, and LightGBoost are four boosting algorithms that were combined with the GWO algorithm to create four different hybrid models. Optimizing and modifying the boosting algorithms’ hyperparameters was the aim of this hybridization. The case study’s findings demonstrated the appropriate accuracy of the suggested methodology for hourly heating energy consumption forecast. The XGBoost-GWO hybrid model has the best evaluation index values in commercial and residential structures, whereas the CatBoost-GWO hybrid model has the best evaluation index values in big venue buildings, according to the research findings. Thus, these models are recommended for predicting heating energy usage on an hourly basis.