CH-24-C087 - Predicting Occupant Thermal Comfort and Local Thermal Discomfort for Multiple HVAC Designs and Weather Scenarios with Data Driven Approach

During technology advancement, the challenge of achieving thermal comfort in big indoor spaces like offices and commercial areas has been met with detailed computational fluid dynamics simulations. Our study introduces an innovative method by combining machine learning with data-driven strategies to simplify this complex task. In our approach, we have implemented UNET Deep Learning Architecture. By considering environmental conditions and spatial information, our research can achieve accurate predictions of thermal comfort and local thermal discomfort in various spaces. Our research also adheres strictly to ASHRAE Standard 55, highlighting our dedication to meeting international standards with focus on energy efficiency. By looking into specific issues of thermal discomfort, we gain a complete understanding of temperature dynamics. Using artificial intelligence alongside these standards, we move past the hurdles of traditional, heavy simulation methods, making the process faster and more cost-effective. To summarize, our work introduces a fresh perspective in HVAC engineering. We provide a solid method for better HVAC design and informed decision-making, all aimed at creating comfortable indoor spaces.