Our poster titled “Appliance Classification and Energy Management Using Multi-Modal Sensing ” won the best poster award at the 3rd ACM Workshop on Embedded Systems for Energy-Efficiency in Buildings (BuidSys 2011) .

The work that we were presenting was based on some preliminary results that we have for the following idea: using data from a variety of sensors in a home (motion, light intensity, plug-level meters, etc.); can we automatically categorize each appliance in the house by finding clusters of correlation between their usage patterns and the activity of the residents?

We found that using a simple statistical correlation as a metric to cluster different signals together, we could determine whether an appliance was “active” (i.e., its usage is highly-correlated with a person being close to it), “passive” (only correlated when the appliance starts) or “background” (not correlated with motion data).

This information is useful for energy management purposes (e.g., reducing energy waste, load shedding and scheduling), and our preliminary results show that it may be possible to determine these categories in an unsupervised fashion.

{ 0 comentarios }