Energy disaggregation techniques for low resolution whole-house energy consumption data

1. A method for learning an energy consumption signature associated with a specific appliance from energy consumption interval data sampled ranging from minutely to hourly associated with a specific house, comprising:

receiving at a processor the energy consumption interval data;

communicating by the processor with a first database comprising non-electrical information;

communicating by the processor with a second database comprising training data, the training data comprised at least in part of:

high resolution energy consumption data, sampled periodically with periods of less than one minute, received from a plurality of homes associated with appliances at least some of which of a same category as the specific appliance; and

feedback information received from a user through an interactive user interface, the feedback information comprises one or more of an indication of whether previous appliance identification was correct and indicating when a specific appliance was used;

determining a correlation between the energy consumption interval data and the training data;

determining based at least in part on the correlation that the specific appliance was running during the acquisition of the interval data; and

identifying the energy consumption signature of the specific appliance.

2. The method of claim 1 , wherein the energy consumption interval data is received from one or more of the group consisting of:

a smart meter;

a meter data management system from a utility;

a website of a utility; and

a non-utility installed sensor or measuring device.

3. The method of claim 1 , wherein the training data further comprises non-electric information comprises home and community specific attributes comprising one or more of the group consisting of:

zip code;

local weather;

home size;

home age;

information about presence of various appliances in the household;

thermostat data;

gas consumption data; and

water consumption data.

4. The method of claim 3 wherein the non-electric information is procured from sources other than the first database or second database.

5. The method of claim 1 , wherein at least some of the training data in the second database is obtained by:

running one or more disaggregation algorithms on the high resolution energy consumption data from the plurality of homes to produce a disaggregated appliance output;

saving the disaggregated appliance output.

6. The method of claim 1 , wherein at least some of the training data in the second database is obtained from user provided information about start and stop times of appliances in the user's home.

7. The method of claim 1 , wherein the step of determining based at least in part on the correlation that the specific appliance was running comprises:

segmenting or grouping training data based upon various criteria;

receive the energy consumption interval data and determine a matching segment or group for analysis;

train a classifier based upon the determined matching segment or group;

use the trained classifier to identify occurrences of usage of the specific appliance in the energy consumption interval data.

8. The method of claim 1 , further comprising detecting heating or cooling appliance consumption in the energy consumption interval data, comprising:

selecting time intervals in which heating or cooling appliances are determined to be running;

applying filters to ensure regularity of the heating and cooling appliance patterns.

9. The method of claim 1 , wherein the step of learning the energy consumption signature of the specific appliance comprises:

identifying any patterns of certain periodic loads in the energy consumption interval data;

determining, based upon non-electrical data, whether the house has a pool;

determining, based on scheduling consistency and the presence of a pool the existence of a pool pump;

determining, based upon the existence of a pool pump and scheduling consistency of certain periodic loads in the energy consumption interval data, the individual energy consumption profile for the pool pump.

10. The method of claim 9 , wherein a confidence level of energy consumption profile for the detected pool pump is calculated based on a number of attributes.