A Framework for Enabling Energy-aware Smart Facilities
Lucio Soibelman, H. Scott Matthews, Jose Moura, Yuanwei Jin, Mario Berges
Left: Electric panel for a residential building with current transformers installed to collect current waveforms. Right: Start-up transient for a refrigerator in the building.
It is no surprise to the average consumer that energy prices, as well as energy consumption, have increased in recent years. However, it is difficult for the conscientious consumer to determine how best to save energy by changing his or her daily activities and use of specific appliances. Building managers and homeowners must typically base their energy consumption decisions on aggregate monthly data, which does not clearly show the contribution of individual appliances. Even though there is adequate evidence to show that consumers can achieve significant energy savings if they are given real-time, appliance-specific data, the available solutions are either inadequately small or prohibitively expensive.
This research attempts to create a framework for obtaining appliance-specific feedback about energy consumption in a building by extracting valuable information from low-cost data sources (e.g., the overall electricity consumption, the time of day, temperature patterns, etc.). Advanced signal processing, data mining and machine learning algorithms are used to obtain appliance-specific electricity data and provide users with recommendations on how to optimize their energy consumption and understand the effects of their energy-related decisions.
A series of residential buildings in Pittsburgh, PA are used to test the electricity monitoring tools. Some of these buildings are operated by a project partner, Blueroof Technologies, a non-profit corporation located in McKeesport, PA that researches, develops, and provides affordable senior-citizen housing with integrated sensor networks and building automation systems. Robert Bosch LLC, a leading global provider of consumer goods and building technology, is another partner and provides additional technical research assistance and expertise.
The main technique used in the approach - Non-Intrusive Load Monitoring - has been a research topic for over two decades, but has not yet produced any widely commercialized products. This work is supported primarily by a National Science Foundation (NSF) grant, which brings an opportunity to explore the main obstacles that have prevented the technology from reaching a wide audience.