Energy Disaggregation

Problem Statement: Build algorithms that take the whole-house power consumption time series from smart meters, and determine which appliance was on at what time in a particular home.

Identify the appliances based on their “fingerprints” on the total power consumption trace, even when multiple appliances running at the same time.
Ability to disaggregate nearly 70% to 80% of the total consumption, by identifying appliances such as HVAC, Cooking, Laundry, Pool Pump, Lights & Fans, Base Load.

Digital Signal Processing, Machine Learning, Bayesian Belief Network, Blind Source Separation, Hidden Markov Models.