If These Appliances Could Talk: Mario Bergés pursues his vision of a smart home
Mario Bergés wants to make our homes smarter. Since he joined CEE as an assistant professor in 2010, Bergés has focused his research efforts on using infrastructure monitoring and machine learning techniques to develop “smart” urban infrastructure -- that is infrastructure that self-regulates in response to changing internal and external conditions. His lab, INFERlab, is currently part of efforts funded by the National Science Foundation, Samsung Electronics, and Hewlett Packard to make buildings active participants in the power grid through monitoring and control of appliances’ energy consumption. This work is done in collaboration with local faculty from CEE (Scott Matthews) and ECE (José M. F. Moura, Anthony Rowe, Bruce Krogh and Soummya Kar), as well as collaborators from University of Southern California and University of Maryland, Eastern Shore.
When an appliance draws power from the grid, it leaves a unique electrical signature that can tell engineers how much power it consumes and whether it requires maintenance. Suman Giri, a PhD candidate in INFERlab, is studying how to disaggregate a home’s total power consumption into individual appliance-level data. With this information at their disposal, consumers could take action by purchasing more energy-efficient appliances or reducing their use of certain appliances. Giri offered a simple analogy: “If you go shopping and receive a final, un-itemized bill, you cannot identify what cost the most or where to reduce your spending. When your bill is disaggregated to the item level, that arms you with the information you need to do something about it.”
Giri’s research on appliance signatures flows naturally into that of his INFERlab colleague and fellow PhD candidate, Emre Can Kara. Kara is working to use appliance-level data to create a central control unit that reduces home energy use in a non-intrusive manner. For instance, some home appliances such as refrigerators can be switched off for short intervals without inconveniencing the user; this results in a small but reliable savings in energy. This concept would then be applied to a collection of appliances across many homes, using them as a single responding unit. Since appliance-level control mechanisms deal with very small amounts of power, they need to take place as a coordinated effort to positively affect the power grid; shutting off a thousand refrigerators for five minutes could make a difference during a power shortage.
Because many consumers balk at the idea of outside control of their appliances, Bergés envisions a compromise: utilities could give consumers the option to take part in a smart appliance program, and anonymize consumer data when it leaves the home. Alternatively, a home energy control device could manage the electricity in a home and exchange minimal information with the utility company. In this scenario, the utility company would alert the control device when there is an excess or shortage of power in the grid, prompting the device to either use energy (e.g. charge a hybrid car) or conserve it (temporarily shut off an appliance). “There is a sense that in the near future, there is going to be a more dynamic balancing act between the utilities, who are providing you power, and the consumers,” Bergés says. “Both a top-down and bottom-up approach to energy conservation are going to be necessary for that to work.”
Adjusting power demand to fit the supply is known as demand response. Though demand response already occurs on a large scale-utilities offer energy-intensive clients a rebate if they restrict their energy use when demand is high-- INFERlab is studying its applicability on a residential scale. Bergés’ research is happening at the right time. While many renewable power sources such as wind and solar have significant advantages, they are intermittent sources, meaning they are not continuously available. As human reliance on intermittent power sources increases, having the option to instantaneously harvest multiple small energy loads when power supply cannot meet consumer demand will play a key role in energy conservation.
The smart home research taking place in INFERlab is only one element of Bergés’ overarching objective. He envisions a self-monitoring system of urban infrastructure that can anticipate and respond to changes in its environment to increase efficiency and robustness. Just as a smart home could adjust its appliances to optimize energy consumption, smart technology can promote savings and detect faults in systems outside of the home as well. Bergés is collaborating with CEE Profs. Irving Oppenheim and James H. Garrett and ECE Profs. José M. F. Moura and David Greve to design self-monitoring pipelines that alert engineers when cracks develop. This technology would increase the independence and safety of fluid transportation systems. In addition, he has joined forces with the U.S. Army Corps of Engineers to design an early warning system for dams and levees.
Bergés wears many hats - researcher, advisor, teacher, even poet - and his dedication to each of these roles is evident. Throughout his work, his students remain his primary focus. “What excites me is coming in here and seeing the students advance and excel,” he explained. “Projects will come and go, but the project itself is nothing compared to the student.” More information about the research being conducted in Bergés’ lab can be found at inferlab.org.