Download

Download Full Text (395 KB)

Description

Heating, Ventilation, and Air Conditioning (HVAC) systems play a vital role in maintaining indoor comfort but are among the largest energy consumers in buildings. Traditional HVAC controllers often operate inefficiently, failing to adapt to changing conditions such as weather variations, occupancy shifts, and electricity pricing. To address this, we introduce STA-DRL (Spatial-Temporal Attention Deep Reinforcement Learning), a framework designed to optimize HVAC performance in multi-zone buildings. STA-DRL combines convolutional neural networks (CNNs) to model heat transfer between rooms, long short-term memory (LSTM) networks to capture time-dependent factors, and an attention mechanism to prioritize energy distribution based on occupancy and cost fluctuations. This allows real-time adjustments to HVAC setpoints, improving efficiency while maintaining occupant comfort. By dynamically balancing energy use and comfort, STA-DRL offers a scalable solution for smart energy management, paving the way for more sustainable and cost-effective building automation. Future research will focus on real-world implementation and integration with smart infrastructure

Publication Date

4-1-2025

Keywords

HVAC Optimization, Deep Reinforcement Learning, Spatial-Temporal Attention, Smart Building Automation

A Spatial-Temporal Attention Deep Reinforcement, A Smart HVAC Control System for Energy Savings and Comfort

Share

COinS