Scholarlink

<p>The increasing need for sustainable and energy-efficient smart buildings has prompted the development of advanced strategies for optimizing real-time occupancy and energy management systems. This study presents an innovative approach that harnesses the power of the Greylag Goose Optimization (GGO) algorithm to enhance deep learning models through effective feature selection and hyperparameter tuning. To address challenges in managing energy consumption while ensuring occupant comfort, we introduce a methodology that combines binary GGO (bGGO) for feature selection with standard GGO for fine-tuning hyperparameters. The proposed method significantly improves the predictive accuracy of occupancy and energy models. Initial results using the baseline Error-Trend-Seasonal (ETS) model yielded a mean squared error (MSE) of 0.00852. After applying bGGO, performance improved markedly, achieving an MSE of 0.000477. Further optimization through GGO hyperparameter tuning reduced the MSE to an exceptional&#xa0;<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(8.23 \times 10^{-7}\)</EquationSource> </InlineEquation>. These outcomes highlight the effectiveness of our framework in improving both accuracy and efficiency in smart building systems. The implications are far-reaching, offering a path toward intelligent, real-time control of HVAC and lighting systems that significantly reduce energy consumption while promoting sustainability and operational efficiency in modern buildings.</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

Greylag Goose Optimization for smart building efficiency: feature selection and hyperparameter tuning in occupancy and energy management

  • 期刊论文

  • 发表日期：

    2026年1月14日

查看全文
  • Amal H. Alharbi,
  • El-Sayed M. El-kenawy,
  • Faris H. Rizk,
  • Khaled Sh. Gaber,
  • Doaa Sami Khafaga,
  • Marwa M. Eid

摘要

The increasing need for sustainable and energy-efficient smart buildings has prompted the development of advanced strategies for optimizing real-time occupancy and energy management systems. This study presents an innovative approach that harnesses the power of the Greylag Goose Optimization (GGO) algorithm to enhance deep learning models through effective feature selection and hyperparameter tuning. To address challenges in managing energy consumption while ensuring occupant comfort, we introduce a methodology that combines binary GGO (bGGO) for feature selection with standard GGO for fine-tuning hyperparameters. The proposed method significantly improves the predictive accuracy of occupancy and energy models. Initial results using the baseline Error-Trend-Seasonal (ETS) model yielded a mean squared error (MSE) of 0.00852. After applying bGGO, performance improved markedly, achieving an MSE of 0.000477. Further optimization through GGO hyperparameter tuning reduced the MSE to an exceptional  \(8.23 \times 10^{-7}\) . These outcomes highlight the effectiveness of our framework in improving both accuracy and efficiency in smart building systems. The implications are far-reaching, offering a path toward intelligent, real-time control of HVAC and lighting systems that significantly reduce energy consumption while promoting sustainability and operational efficiency in modern buildings.