Download Full Text (403 KB)
This research addresses key issues for applying advanced building data analytics to energy efficient control opportunities. First the research identifies advancements and potential hurdles around the three primary means for acquiring data: energy management systems, dedicated measurement systems, and advanced computer software that accesses and archives data from energy management systems. These are described using case studies from commercial building control systems and web-based real time dedicated measurement technology. Next, the research describes effective rule-based data analytics and control strategies that are traditionally used. Rule-based data analytics utilize specific knowledge about HVAC systems to identify key data points and analytical methods to identify energy saving opportunities and develop improved control algorithms. The research describes both theory and application of these rule-based analytics for the control of systems like air-side economizer, ventilation fans, pumping and chilled water systems. Finally, the research proposes a framework to apply advanced machine learning and data mining techniques to the same problem. Machine-learning control differs from rule-based control in that this control type requires less specific knowledge about HVAC systems. The proposed framework uses existing data, where available, to pattern match and build robust models emulating the performance of the system under consideration. To these models, classical optimization algorithms (knapsack, greedy and shortest distance) and mathematical framework (Game theory and Design of Experiments) are adapted and applied to reach the best control strategy. For systems without past performance data, a stochastic framework using decision chains (Markov processes) and adaptive controls using the reinforcement learning method is proposed for the same. These techniques are demonstrated on select systems e.g. Pumping plants and HVAC systems.
J Kelly Kissock
Primary Advisor's Department
Mechanical and Aerospace Engineering
Stander Symposium poster
"Advanced Data Analytics and Optimal Control of Building Energy Systems" (2019). Stander Symposium Posters. 1611.