Learning power grid characteristics to anticipate load

Patent metadata

Field	Value
Publication number	US-10628762-B2
Application number	US-201815948750-A
Country	US
Kind code	B2
Filing date	Apr 9, 2018
Priority date	Apr 9, 2018
Publication date	Apr 21, 2020
Grant date	Apr 21, 2020

Abstract

Improving the operations of a data center that is located within a power grid and that includes its own power sources. Past operational characteristics of the power grid are analyzed to derive learned characteristics for the power grid. Current and/or future operational characteristics of the power grid are also monitored. A prediction regarding an upcoming, anticipated load for the power grid is then generated using the learned characteristics and the monitored characteristics. In response to this prediction, one or more mitigation operations are selected and then performed at the data center to ensure that the data center is adequately available. Some of these mitigation operations include, but are not limited to, causing the data center to consume more power, causing the data center's power sources to store more power, or causing the data center to migrate services and/or data to a different data center.

First claim

What is claimed is:

1. A computer system comprising:
one or more processors; and
one or more computer-readable hardware storage devices having stored thereon computer-executable instructions that are structured to be executable by the one or more processors to thereby cause the computer system to:
analyze past operational characteristics for a power grid within which a data center is currently operating to derive learned characteristics for the power grid, the data center including one or more power sources that are structured to provide power to the data center independently of the power grid;
identify one or more possible future operational characteristics for the power grid;
use the learned characteristics and the identified future operational characteristics to generate a prediction regarding a future load on the power grid, the prediction including a buffer selected to compensate for at least some unpredictability associated with the future load of the power grid; and
in response to the prediction, perform one or more mitigation operations at the data center to ensure a determined availability of services associated with the data center, the one or more mitigation operations including at least one of:
causing the data center to consume more power;
causing the one or more power sources to store power; or
migrating services from the data center to a different data center.

2. The computer system of claim 1 , wherein the one or more power sources includes a fuel cell.

3. The computer system of claim 1 , wherein the one or more mitigation operations includes causing the data center to consume more power, and wherein consuming more power is performed by increasing a load for a climate control system of the data center.

4. The computer system of claim 1 , wherein performing the one or more mitigation operations is based, at least partially, on a service level agreement.

5. The computer system of claim 1 , wherein deriving the learned characteristics for the power grid includes analyzing past environmental conditions that have occurred in geographic areas associated with the power grid.

6. The computer system of claim 1 , wherein the learned characteristics include information corresponding to a load of the power grid, and wherein the information includes one or more of timing metrics for the power grid, cost metrics for the power grid, location data for the power grid, past political events that have impacted the power grid, or past social events that have impacted the power grid.

7. The computer system of claim 1 , wherein the buffer is a pre-determined buffer.

8. The computer system of claim 1 , wherein generating the prediction regarding the future load on the power grid includes generating a level of confidence associated with the prediction, the level of confidence indicating an estimated likelihood that the future load will actually occur.

9. The computer system of claim 1 , wherein the one or more mitigation operations are based, at least partially, on a power generation ability of the data center, a power storage ability of the data center, or a power consumption ability of the data center.

10. The computer system of claim 1 , wherein the one or more mitigation operations are performed within a threshold number of milliseconds of a power grid fluctuation.

11. The computer system of claim 1 , wherein analyzing the past operational characteristics for the power grid includes monitoring fluctuations in a load of the power grid during different times of a day or during different times of a year.

12. A method for operating an architecture that improves data center operations, the method being performed by a computer system that operates within the architecture, the method comprising:
analyzing past operational characteristics for a power grid within which a data center is currently operating to derive learned characteristics for the power grid, the data center including one or more power sources that are structured to provide power to the data center independently of the power grid;
identifying one or more possible future operational characteristics for the power grid;
using the learned characteristics and the identified future operational characteristics to generate a prediction regarding a future load on the power grid, the prediction including a buffer selected to compensate for at least some unpredictability associated with the future load of the power grid; and
in response to the prediction, performing one or more mitigation operations at the data center to ensure a determined availability of services associated with the data center, the one or more mitigation operations including at least one of:
causing the data center to consume more power;
causing the one or more power sources to store power; or
migrating services from the data center to a different data center.

13. The method of claim 12 , wherein the one or more mitigation operations includes migrating the services from the data center to the different data center.

14. The method of claim 12 , wherein analyzing the past operational characteristics for the power grid includes analyzing weather data for a geographic region serviced by the power grid.

15. The method of claim 12 , wherein the one or more mitigation operations includes migrating the services from the data center to the different data center, and wherein the different data center is using a different power grid.

16. The method of claim 12 , wherein performing the one or more mitigation operations is performed after generating a decision tree that takes in as input the prediction.

17. The method of claim 12 , wherein deriving the learned characteristics is performed using a neural network, and wherein analyzing the past operational characteristics is performed by the neural network.

18. The method of claim 12 , wherein performing the one or more mitigation operations is performed after generating a decision tree that takes in as input (1) the prediction and (2) a latency between the data center and at least one other data center.

19. The method of claim 12 , wherein performing the one or more mitigation operations includes causing the data center to consume more power, and wherein determining how much more power to consume is based, at least partially, on a determined mean time to failure associated with one or more hardware devices of the data center.

20. One or more hardware storage devices having stored thereon computer-executable instructions that are structured to be executable by one or more processors of a computer system to thereby cause the computer system to:
analyze past operational characteristics for a power grid within which a data center is currently operating to derive learned characteristics for the power grid, the data center including one or more power sources that are structured to provide power to the data center independently of the power grid;
identify one or more future operational characteristics for the power grid;
use the learned characteristics and the identified future operational characteristics to generate a prediction regarding a future load on the power grid, the prediction including a buffer selected to compensate for at least some unpredictability associated with the future load of the power grid; and
in response to the prediction, perform one or more mitigation operations at the data center to ensure a determined availability of services associated with the data center, the one or more mitigation operations including at least one of:
causing th

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Helsel Michael Rees
• Blas Vincent Joseph
• Keehn Nicholas Andrew
• James Sean Michael
• Middaugh Brandon

Classifications

• G05F1/66

  Regulating electric power · CPC title

• G06Q10/06312

  Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling · CPC title

• G06Q10/04 Primary

  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 title

• G06Q50/06

  Energy or water supply · CPC title

• G05B15/02

  electric · CPC title