Interesting talk at the Bloomington Data Collective Meetup last night by Zach Dwiel of Plot Watt. Plot Watt uses energy sensors in households and restaurants to find opportunities to save energy. In restaurants, the use cases might be
-discovering that the restaurant’s outdoor lights stay on all night despite the restaurant closing at 10pm
-noticing when the fryer is left on after closing and alerting management
-realizing that the appliances in use are much less energy efficient than similar appliances at other locations
-determining poor installs that are costing energy
-finding poorly specified defaults on the HVAC or refrigeration units.

Zach is part of the software team that is in charge of dis-entangling the differnet signals in the electric load to determine when specific appliances/lights/HVAC are clicking on/off. This is a very difficult problem, and the academic literature has been limited by solid “truth” examples from which to learn/train. Zach discussed how Plot Watt is using machine learning to try to get at the problem. They may have the most data on this problem of anyone.

Related: It looks like a startup may be trying to automate this approach and bring it to more households.

Thanks to Crisson for the invite to the meetup.