Self-learning, real-time, data-driven power metering system

• US 9,705,327 B2
• Filed: 02/12/2016
• Issued: 07/11/2017
• Est. Priority Date: 07/14/2015
• Status: Active Grant

• Current Assignee: REstore NV (Centrica PLC)
• Original Assignee: REstore NV (Centrica PLC)
• Inventors: Rombouts, Jan-Willem
• Primary Examiner(s): Everett, Christopher E
• Assistant Examiner(s): Pan, Yuhui R
• Application Number:
  US15/043,006
• Publication Number:
  US 20170018923A1
• Time in Patent Office: 515 Days
• Field of Search: 700295
• CPC Class Codes: G05B 17/02 electric
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 50/06 Energy or water supply
• CPC Class Codes: H02J 3/00 Circuit arrangements for ac mains or ac distribution networks
• CPC Class Codes: Y02P 80/10 Efficient use of energy, e.g. using compressed air or pressurized fluid as energy carrier

Abstract

A power metering system includes any number of sites and a central database. Loads within a site are unmetered and a sensor data vector over time includes sensor data, operational data and external data. Iteration occurs over time intervals for all unmetered loads and the unmeasured power for each interval is disaggregated by matching a load type with a computer model in the central database having the same feature domain in order to predict the power usage of each load. Iteration again occurs over those intervals in which only a single load is operating; the power of that load is determined to be the unmeasured power minus miscellaneous power, and a new computer model is fitted for that single load and also uploaded to the database. The feature domain of an existing model may also be increased. Both iteration steps repeat until no more unmeasured power can be disaggregated.

Claims

1. A method of disaggregating power at a site, said method comprising:

identifying a first time interval when only a single first load is operating at said site;

determining a first load power used during said first time interval by said first load;

fitting a model using selected sensor data of said site during said first time interval and said determined first load power;

determining a second load power used by said first load during a second time interval during which at least one other second load is operating using said model;

outputting said first and second load powers;

identifying one of a plurality of loads at said site for which, during a time longer than for other loads of said plurality of loads, no model can be fit while said one load is operating, said one load not including a power meter; and

adding a power meter to said one load to measure a load power used by said one load.

2. A method as recited in claim 1 further comprising:

determining said first load power by subtracting miscellaneous power of said site from an unmeasured power of said entire site during said first time interval.

3. A method as recited in claim 1, wherein only said first load and said second load are operating during said second time interval, said method further comprising:

subtracting said second load power and a miscellaneous power from an unmeasured power of said site during said second time interval in order to obtain a load power used by said second load during said second time interval; and

outputting said load power used by said second load during said second time interval.

4. A method as recited in claim 1 further comprising:

determining a load power used by a third load during a third time interval at another site different from said site during which at least one other load is operating by inputting features of said third load into said model to produce said load power, said third load being of the same type as said first load; and

outputting said load power used by said third load during said third time interval.

5. A method as recited in claim 4 further comprising:

uploading said model and a domain of said selected sensor data to a global model database such that said model and said domain are accessible by both said site and said another site.

6. A method as recited in claim 1 wherein said first load does not include an individual power meter.

7. A method as recited in claim 4 wherein said third load does not include an individual power meter.

8. A method as recited in claim 4 wherein said site and said another site are controlled by different entities.

9. A method as recited in claim 1 further comprising:

computing a power disaggregation error for said site and identifying one of the plurality of loads at said site that contributes most to said disaggregation error, said one load not including a power meter;

adding a power meter to said one load that contributes most to said disaggregation error in order to measure a load power used by said one load.

10. A method as recited in claim 1 further comprising:

ranking a plurality of loads at said site from worst to best based upon a length of time during which each of said loads is operating but no model is available; and

adding a power meter to one of said ranked loads to measure a load power used by said one load.

11. A method of disaggregating power of a site, said method comprising:

out of a plurality of loads operating at said site during a plurality of time intervals, identifying a model in a database that predicts power used by a first of said loads during one of said time intervals;

subtracting said predicted power from a running total of unmeasured power used by said site during said time interval and indicating that said first load is not operating during said time interval;

determining power used by a second of said loads during one of said time intervals when no other loads are operating or are indicated as operating;

fitting a new model using said determined power and storing said new model in said database;

outputting said predicted power and said determined power;

identifying a third of said plurality of loads at said site for which, during a time longer than for other loads of said plurality of loads, no model can be fit while said third load is operating, said third load not including a power meter; and

adding a power meter to said third load to measure a load power used by said third load.

12. A method as recited in claim 11 further comprising:

iterating over said steps of claim 11 until no more outputting can be performed.

13. The method as recited in claim 11 further comprising:

determining said determined power by subtracting miscellaneous power of said site from an unmeasured power of said entire site during one of said time intervals when no other loads are operating or are indicated as operating.

14. A method as recited in claim 12 further comprising:

determining a load power used by a fourth load during a time interval at another site different from said site during which at least one other load is operating by inputting features of said fourth load into said new model fitted at said site to produce said load power; and

outputting said load power used by said fourth load during said time interval.

15. A method as recited in claim 14 further comprising:

uploading said model and a domain of selected sensor data to a global model database such that said model and said domain are accessible by both said site and said another site.

16. A method as recited in claim 11 wherein at least some of said loads do not include individual power meters.

17. A method as recited in claim 16 wherein none of said loads include individual power meters before said identifying.

18. A method as recited in claim 12 further comprising:

computing a power disaggregation error for said site and identifying one of said plurality of loads at said site that contributes most to said disaggregation error, said identified one load not including a power meter;

adding a power meter to said identified one load that contributes most to said disaggregation error in order to measure a load power used by said identified one load.

19. A method of disaggregating power at a plurality of sites, said method comprising:

identifying a first time interval when only a single first load is operating at a first site;

determining a first load power used during said first time interval by said first load;

fitting a model using selected sensor data of said first site during said first time interval and said determined first load power;

identifying a second load operating at a second site different from said first site and being of the same type as said first load operating at said first site;

determining a second load power used during a second time interval by said second load using said model;

outputting said determined second load power;

identifying one of a plurality of loads at said first site for which, during a time longer than for other loads of said plurality of loads, no model can be fit while said one load is operating, said one load not including a power meter; and

adding a power meter to said one load to measure a load power used by said one load.

20. A method as recited in claim 19 wherein said first site and said second site are controlled by different entities.

21. A method as recited in claim 19 wherein loads other than said second load are also operating during said second time interval at said second site.

22. A method as recited in claim 19 further comprising:

uploading said model from said first site to a central database; and

determining said second load power using said model in said central database.

23. A method as recited in claim 19 further comprising:

determining said second load power when feature values of said second load fall within a feature space of the selected sensor data used to fit said model.

24. A method as recited in claim 19 wherein said first and second loads do not have individual power meters.

25. A method as recited in claim 19 further comprising:

computing a power disaggregation error for said first site and identifying one of a plurality of loads at said first site that contributes most to said disaggregation error, said one load not including a power meter;

adding a power meter to said one load that contributes most in order to measure a load power used by said one load.