Back to All Events

Chenlu Zhang PhD Proposal - MMCH 415 IW

  • CMU School of Architecture - MMCH 415 IW (map)

Human in the loop: A bio-sensing and agent-based approach for personalized thermal comfort controls in offices
PhD Proposal by Chenlu Zhang, PhDBPD Candidate

Advisory committee:
Vivian Loftness (Chair), University Professor
Sivaraman Balakrishnan, Assistant Professor
Ömer Tugrul Karaguzel, Assistant Teaching Professor
Azizan Abdul-Aziz, Assistant Research Professor
Bertrand Lasternas(External), Associate Director


A comfortable indoor thermal environment plays a crucial role in preserving occupant health and productivity. Today, in most office buildings, thermal comfort is regulated only by indoor temperature and schedule-based rules. While prevalent, this control strategy has resulted in low thermal satisfaction rates. A growing number of researchers are focusing on the development of personalized thermal comfort models through feedback to establish individual-specific thermal comfort requirements for occupant-centric controls. However, the strengths of these personalized thermal comfort models are compromised for negotiated control in shared multi-occupant spaces. Therefore, this thesis proposes an agent-based framework for personalized thermal comfort control to place the human-in-the-loop by continuously integrating occupant bio-signals, subjective feedback and environmental conditions into the operation of different heating and cooling systems.

The agent-based framework is built upon personalized thermal comfort models and control agents, that can flexibly represent the individuals in negotiated controls. The personalized control agent uses a reinforcement learning approach for optimal control and learning through interaction. It establishes individual control preferences related to thermal comfort to initiate the best control actions on behalf of an occupant, given the constraints of the building systems. The negotiations of preferred control actions between occupants will maximize collective comfort and minimize energy. 

The proposed framework integrated bio-sensing and reinforcement learning agents for personalized thermal comfort control in offices will contribute to improvements in controlling building systems that deliver occupant thermal comfort, with benefits for health and productivity. The human in the loop research will address the range of controllability options from personalized settings in individual occupant spaces to negotiated settings in multi-occupant spaces.

View the PhD abstract and PhD proposal at these links.