Data-driven smart buildings
Smart buildings are at the intersection of three key pillars: building, occupant, and aim for occupant-centric, grid-interactive and energy-efficient performance. There are four steps to achieve smart buildings: sensing, prediction, modelling, and optimization. Data-driven approaches could be applied in and assist all of the four steps mentioned above. In this seminar, I will introduce our research on how to apply machine learning techniques to enhance the accuracy, cost-effectiveness and scalability of sensing, to improve prediction accuracy, and to fill in the performance gap between simulation and reality. Last, I will discuss why data-driven approaches are favored in certain building-related studies and applications.
Dr. Zhe Wang is a postdoc at BTUS Division. His research applies machine learning tecniques to model predictive control and occupant behavior modeling. Before joining LBNL, he worked as an energy consultant at the World Bank, and as a postdoc with CBE of University of California Berkeley on personalized thermal comfort model and devices. Dr. Zhe Wang received his Bacheor degree and Ph.D. degree from Tsinghua University (2017) with a focus on intermittent heating for residential buildings. He also received a Master degree from University of Cambridge (2014) with a focus on experimental fluid dynamics.