Non-Intrusive Interpretation and Improvement of Multi-Occupancy Human Thermal Comfort through Analysis of Facial Infrared Thermography

在美国和全世界，暖通空调系统是最大的能源终端用途之一，约占运营住宅和商业建筑所需总能源的50%。尽管暖通空调系统有很大的能源消耗，但居住在建筑环境中的人往往对自己的热舒适性不满意。目前的“人在回路”方法为居住者提供了对热舒适性和偏好进行投票的机会(例如，通过调整恒温器)，从而允许根据人的反馈来调整暖通空调系统。然而，依赖于间歇性的人类反馈，阻碍了对舒适水平的可靠评估和舒适设置点的确定。该项目将探索在单人和多人居住的建筑空间中使用红外热像仪作为一种非侵入性方法预测人体热舒适偏好的可行性。该项目还将设计和验证建筑物的稳健的暖通空调控制框架，该框架将同步使用分析的热像数据来调整其设置点，以提高室内空间的热舒适性，并减少多人空间中居住者的总体不满。该项目旨在探索在真实的、可操作的建成环境中非侵入性和可靠地测量热舒适性的前提。由此产生的新知识有可能将建筑暖通空调控制从被动的、用户经验的过程转变为自动的、以用户为中心的、数据驱动的机制，可以同时提高室内环境中的居住者满意度，同时降低能源消耗。该研究从人体热舒适性评价、计算机视觉和不确定条件下的优化等方面扩展了理论，旨在创建一种健壮的、可扩展的非侵入式暖通空调控制框架，用于各种室内环境下的热舒适性优化。虽然本项目的重点是热舒适性，但开发的框架和方法可以扩展到评估其他室内环境质量因素，如照明和气流。该项目还将建立一个公开可用的热舒适数据集，包括面部热图像、居住者的主观热感觉和偏好以及周围房间条件，这将使未来能够对新方法进行一致的评估和基准。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

In the U.S. and worldwide, HVAC systems represent one of the largest energy end uses, accounting for approximately 50 percent of the total energy required to operate residential and commercial buildings. Despite the significant energy footprint of HVAC systems, occupants in the built environment are often dissatisfied with their thermal comfort. Current "human-in-the-loop" approaches provide opportunities for occupants to vote on thermal comfort and preferences (for example, by adjusting the thermostat), thus allowing for HVAC system adjustment based on human feedback. However, relying on intermittent human feedback prevents robust evaluation of the comfort level and determination of a comfortable setpoint. This project will explore the feasibility of using infrared thermography as a non-intrusive method for predicting human thermal comfort preferences in single and multi-occupancy building spaces. It will also design and validate a robust HVAC control framework for buildings that will synchronously use the analyzed thermography data to adjust its setpoint to improve thermal comfort in indoor spaces and reduce overall dissatisfaction among occupants in multi-occupancy spaces.The project aims to explore the premise that thermal comfort can be measured non-intrusively and reliably in real, operational built environments. The resulting new knowledge has the potential to transition building HVAC control from a passive and user-empirical process to an automated, user-centric and data-driven mechanism that can simultaneously improve occupant satisfaction in indoor environments while reducing energy consumption. The research extends theory from human thermal comfort evaluation, computer vision and optimization under uncertainties with the objective of creating a robust and scalable non-intrusive HVAC control framework for thermal comfort optimization in various indoor contexts. Although the focus is on thermal comfort in this project, the developed framework and methodology can be extended to evaluate other indoor environmental quality factors such as lighting and airflow. This project will also build a publicly available thermal comfort dataset consisting of facial thermal images, subjective thermal sensations and preferences of occupants, and ambient room conditions, that will enable consistent evaluation and benchmarking of new methods in the future.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.