Experimental Evaluation of Data-Driven Predictive Indoor Thermal Management

This paper considers the problem of thermal management in a typical shared indoor space that may be equipped with multiple heterogeneous heat sources and have different temperature requirements in different sections (thermal zones) of the shared space. Utilizing an on-campus smart conference room as a testbed, we discuss the practical challenges involved in real-time data-driven model learning, when a simple first-order dynamical model is used to capture the dependencies between the heat controls and the air temperatures measured at sensor locations. The data-driven model is then utilized for predictive control of the thermal environment towards minimizing the error between the desired and attained temperatures, and the integrated solution is evaluated against a standard thermal control employed by the BMS.

[1]  Elie Azar,et al.  Agent-Based Modeling of Occupants and Their Impact on Energy Use in Commercial Buildings , 2012, J. Comput. Civ. Eng..

[2]  Koushik Kar,et al.  Singular Perturbation Method for Smart Building Temperature Control Using Occupant Feedback , 2018 .

[3]  Koushik Kar,et al.  BEES: Real-time occupant feedback and environmental learning framework for collaborative thermal management in multi-zone, multi-occupant buildings , 2016 .

[4]  John T. Wen,et al.  Office building model identification and control design , 2014, 2014 American Control Conference.

[5]  Koushik Kar,et al.  Incentive-Based Mechanism for Truthful Occupant Comfort Feedback in Human-in-the-Loop Building Thermal Management , 2018, IEEE Systems Journal.

[6]  John E. Taylor,et al.  Assessing Eco-Feedback Interface Usage and Design to Drive Energy Efficiency in Buildings , 2012 .

[7]  Koushik Kar,et al.  Collaborative Energy and Thermal Comfort Management Through Distributed Consensus Algorithms , 2015, IEEE Transactions on Automation Science and Engineering.

[8]  Milind Tambe,et al.  Coordinating occupant behavior for building energy and comfort management using multi-agent systems , 2012 .

[9]  Alberto Cerpa,et al.  Thermovote: participatory sensing for efficient building HVAC conditioning , 2012, BuildSys@SenSys.

[10]  John T. Wen,et al.  Design and instrumentation of an intelligent building testbed , 2015, 2015 IEEE International Conference on Automation Science and Engineering (CASE).

[11]  Rahul Mangharam,et al.  Data Predictive Control for building energy management , 2017, 2017 American Control Conference (ACC).