Events login

Main Events Calendar

Warning Icon This event is in the past.
February 13, 2018 | 11:30 a.m. - 12:20 p.m.
Category: Seminar
Location: State Hall #101 | Map
5143 Cass
Detroit, MI 48202
Cost: Free
Audience: Academic Staff, Alumni, Community, Current Graduate Students, Current Undergraduate Students, Faculty, Parents, Prospective Students, Staff

Black-Box Solar Disaggregation and its Privacy Implication


The aggregate solar capacity in the U. S. is rising rapidly due to continuing decreases in the cost of solar modules.  This increasing solar penetration is imposing operational challenges on utilities in balancing electricity's real-time supply and demand.  To address these issues, both academia and industry have a strong interest in developing solar data analytics to accurately monitor, predict, and react to variations in intermittent solar power. In this talk, I present multiple novel solar analytics using a new hybrid black-box approach that combines "white-box" physical models of solar generation with sophisticated "black-box" machine learning (ML) techniques. Unlike prior approaches, our approach enables a wide range of accurate solar analytics, including solar performance modeling, solar disaggregation, and solar localization, using limited training data and without knowledge of key system parameters. In this talk, I will leverage this model to disaggregate solar generation from energy consumption in "net" meter data that combines them, which both provides utilities new visibility into grid solar production and makes decades of prior work on analytics for energy consumption data applicable to net meter data.  However, while our solar analytics demonstrate how big data techniques can benefit grid operations, I also show that they can have serious privacy implications. In particular, I identify a new privacy threat in energy data: even when anonymized, it embeds detailed information about its location.  I present techniques for extracting this location using both solar and weather signatures embedded in the data, and then discuss the implications of this threat and potential ways to mitigate it. 


Dong Chen is a Ph.D. Candidate in the Department of Electrical and Computer Engineering at the University of Massachusetts Amherst. His research focuses broadly on big data analytics, cybersecurity, and privacy in the context of cyber-physical systems, such as the electric grid. He received an M.S. and Ph.D. from Northeastern University (China) in 2010 and 2014, respectively, and a B.S. in Computer Science from Xi'an Communications Institute in 2006. His recent research focuses on both developing a wide range of energy data analytics for smart meter and renewable energy data and designing techniques for strengthening the security and data privacy of cyber-physical systems. He has published over 20 research papers in these and related areas over the past decade. In addition, his work on energy analytics has been applied to real-world energy data, which he has made publicly available.  For example, his recent work on Non-Intrusive Occupancy Monitoring and the associated data has been cited over 70 times. 

For more information about this event, please contact LaNita Stewart at 313-577-2478 or