Forecasting solar power output

Patent metadata

Field	Value
Publication number	US-10732319-B2
Application number	US-201715690312-A
Country	US
Kind code	B2
Filing date	Aug 30, 2017
Priority date	Aug 30, 2017
Publication date	Aug 4, 2020
Grant date	Aug 4, 2020

Abstract

A method, computer system, and computer program product. Weather forecast data is generated with respect to an area encompassing a location of a solar farm by a computer system. Solar power output by the solar farm is forecasted by the computer system based on the generated weather forecast data. Forecasted solar power output data is generated by the computer system based on the forecasted solar power output by the solar farm. A power grid operation, including one or both of a power grid balancing operation and a power grid optimization operation, is performed based on the forecasted solar power output data.

First claim

What is claimed is:

1. A computer-implemented method comprising:
generating weather forecast data with respect to an area encompassing a location of a solar farm;
forecasting solar power output by the solar farm based on the generated weather forecast data and a deep learning model based on a volumetric convolution neural network;
generating forecasted solar power output data based on the forecasted solar power output by the solar farm; and
performing a power grid operation based on the forecasted solar power output data, wherein the power grid operation is selected from a group consisting of a power grid balancing operation and a power grid optimization operation.

2. The computer-implemented method of claim 1 , wherein the generated weather forecast data comprises data relating to one or more conditions selected from a group consisting of incident solar irradiance, temperature, humidity, pressure, moisture, wind velocity, cloud coverage, visibility, precipitation type and intensity, and aerosol concentrations.

3. The computer-implemented method of claim 1 , wherein the deep learning model is further based on a recurrent neural network.

4. The computer-implemented method of claim 1 , wherein the volumetric convolution neural network comprises at least 5 volumetric convolution layers, at least 3 volumetric max pooling layers, and at least 2 fully connected layers.

5. The computer-implemented method of claim 4 , wherein each of the at least 5 volumetric convolution layers further comprise a Rectified Linear Units layer.

6. The computer-implemented method of claim 5 , wherein the volumetric convolution neural network further comprises an identity skip connection between the layers.

7. A computer system comprising:
one or more computer processors, one or more computer-readable storage media, and program instructions stored on one or more of the computer-readable storage media for execution by at least one of the one or more computer processors, the program instructions, when executed by the at least one of the one or more computer processors, causing the computer system to perform a method comprising:
generating weather forecast data with respect to an area encompassing a location of a solar farm;
forecasting solar power output by the solar farm based on the generated weather forecast data and a deep learning model based on a volumetric convolution neural network;
generating forecasted solar power output data based on the forecasted solar power output by the solar farm; and
performing a power grid operation based on the forecasted solar power output data, wherein the power grid operation is selected from a group consisting of a power grid balancing operation and a power grid optimization operation.

8. The computer system of claim 7 , wherein the generated weather forecast data comprises data relating to one or more conditions selected from a group consisting of incident solar irradiance, temperature, humidity, pressure, moisture, wind velocity, cloud coverage, visibility, precipitation type and intensity, and aerosol concentrations.

9. The computer system of claim 7 , wherein the deep learning model is further based on a recurrent neural network.

10. The computer system of claim 7 , wherein the volumetric convolution neural network comprises at least 5 volumetric convolution layers, at least 3 volumetric max pooling layers, and at least 2 fully connected layers.

11. The computer system of claim 10 , wherein each of the at least 5 volumetric convolution layers further comprise a Rectified Linear Units layer.

12. The computer system of claim 11 , wherein the volumetric convolution neural network further comprises an identity skip connection between the layers.

13. A computer program product comprising:
One or more non-transitory computer-readable storage devices and program instructions stored on at least one or more non-transitory computer-readable storage devices for execution by at least one or more computer processors of a computer system, the program instructions, when executed by the at least one of the one or more computer processors, causing the computer system to perform a method comprising:
generating weather forecast data with respect to an area encompassing a location of a solar farm;
forecasting solar power output by the solar farm based on the generated weather forecast data and a deep learning model based on a volumetric convolution neural network;
generating forecasted solar power output data based on the forecasted solar power output by the solar farm; and
performing a power grid operation based on the forecasted solar power output data, wherein the power grid operation is selected from a group consisting of a power grid balancing operation and a power grid optimization operation.

14. The computer program product of claim 13 , wherein the generated weather forecast data comprises data relating to one or more conditions selected from a group consisting of incident solar irradiance, temperature, humidity, pressure, moisture, wind velocity, cloud coverage, visibility, precipitation type and intensity, and aerosol concentrations.

15. The computer program product of claim 13 , wherein the deep learning model is further based on a recurrent neural network.

16. The computer program product of claim 13 , wherein the volumetric convolution neural network comprises at least 5 volumetric convolution layers, at least 3 volumetric max pooling layers, and at least 2 fully connected layers.

17. The computer program product of claim 16 , wherein each of the at least 5 volumetric convolution layers further comprise a Rectified Linear Units layer.

Assignees

• IBM

Inventors

• Feng Minwei
• Khabibrakhmanov Ildar
• Kumar Tarun
• Lavin Mark A
• Warren Kevin W
• Zhang Rui
• Zhang Wei

Classifications

• H02J2101/24

  Photovoltaics · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/048

  Activation functions · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title