Scholarlink

<p>This study details the design and implementation of an intelligent HVAC optimization system in the Shenzhen Qianhai Smart Community, utilizing advanced machine learning methods like Graph Attention Networks (GATs) and stacking ensemble learning. A comprehensive sensor network monitored temperature, humidity, occupancy, and air quality, allowing for real-time data collection and responsive control. Data preprocessing involved Z-score normalization and feature engineering to improve model accuracy. The system employed Graph construction based on Pearson Correlation Coefficients, resulting in quality embeddings for the GATs. The stacking ensemble combined Gradient Boosting Machines, Neural Networks, and Random Forests, achieving a high Area Under the Curve (AUC) of 0.93. The deployment led to a 15% reduction in energy consumption and an increase in occupant satisfaction. Comparative analysis shows the strength of the GATs and ensemble learning approach over existing systems. This case study validates the methodology and presents a scalable model for energy optimization in smart urban settings. Future work will focus on expanding the system to more communities, integrating renewable energy, and improving real-time capabilities with reinforcement learning.</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

Enhancing Intelligent HVAC optimization with graph attention networks and stacking ensemble learning, a recommender system approach in Shenzhen Qianhai Smart Community

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年2月11日

查看全文
  • Yuan He,
  • Ali B. M. Ali,
  • Saman Ahmad Aminian,
  • Kamal Sharma,
  • Saurav Dixit,
  • Sakshi Sobti,
  • Rifaqat Ali,
  • M. Ahemedei,
  • Husam Rajab,
  • Maryam Alsadat Ziaei Mazinan

摘要

This study details the design and implementation of an intelligent HVAC optimization system in the Shenzhen Qianhai Smart Community, utilizing advanced machine learning methods like Graph Attention Networks (GATs) and stacking ensemble learning. A comprehensive sensor network monitored temperature, humidity, occupancy, and air quality, allowing for real-time data collection and responsive control. Data preprocessing involved Z-score normalization and feature engineering to improve model accuracy. The system employed Graph construction based on Pearson Correlation Coefficients, resulting in quality embeddings for the GATs. The stacking ensemble combined Gradient Boosting Machines, Neural Networks, and Random Forests, achieving a high Area Under the Curve (AUC) of 0.93. The deployment led to a 15% reduction in energy consumption and an increase in occupant satisfaction. Comparative analysis shows the strength of the GATs and ensemble learning approach over existing systems. This case study validates the methodology and presents a scalable model for energy optimization in smart urban settings. Future work will focus on expanding the system to more communities, integrating renewable energy, and improving real-time capabilities with reinforcement learning.