Operating a solar power generating system

Patent metadata

Field	Value
Publication number	US-10103548-B2
Application number	US-201514921988-A
Country	US
Kind code	B2
Filing date	Oct 23, 2015
Priority date	Oct 23, 2015
Publication date	Oct 16, 2018
Grant date	Oct 16, 2018

Abstract

According to some embodiments, the present disclosure may include a method of analyzing solar power forecasts that may include obtaining a test dataset of historical irradiance at a location of a solar power generating system, and normalizing the test dataset based on a clear sky model at the location. The method may also include clustering the test dataset into multiple weather classes that each include a set of characteristics, obtaining a forecast of irradiance at the solar power generating system, and classifying the forecast into one of the weather classes, and determining confidence intervals of the forecast based on the set of characteristics of the one of the plurality of weather classes. The method may additionally include, based on the confidence intervals of the forecast, performing one of increasing output or decreasing output of a source of energy alternative to solar energy.

First claim

What is claimed is:

1. A method of evaluating solar energy forecasts, the method comprising:
generating power at a virtual power plant, the virtual power plant including at least one solar power generating device;
obtaining a test dataset of historical irradiance at a location of the solar power generating device;
normalizing the test dataset by dividing a point in the test dataset at a given time by a clear sky irradiance at the given time;
plotting a histogram of the test dataset;
fitting a curve to the histogram, the curve including a combination of class-specific curves for each of a plurality of classes;
extracting the class-specific curves for each of the classes;
clustering each point in the test dataset with one of the classes based on how closely each point fits to the class-specific curves;
obtaining a forecast of irradiance for a time at the solar power generating device;
finding a nearest neighbor of the forecast in the test dataset based on similarities between the forecast and the test dataset, the nearest neighbor including a point in the test dataset with similar weather conditions to the forecast;
based on the nearest neighbor of the forecast, classifying the forecast into one of the plurality of classes to which the nearest neighbor belongs;
determining confidence intervals of the forecast based on a degree of unpredictability of irradiance of the one of the plurality of classes; and
based on the confidence intervals, modifying power production at the virtual power plant by performing one of increasing output or decreasing output of the solar power generating device based on the confidence intervals.

2. The method of claim 1 , wherein the confidence intervals are determined using a Kriging model.

3. The method of claim 1 , wherein the virtual power plant further includes at least one non-solar power generating device, and modifying the power production at the virtual power plant includes performing one of:
increasing output of the solar power generating device and decreasing output of the non-solar power generating device proportional to the confidence intervals; and
decreasing output of the solar power generating device and increasing output of the non-solar power generating device proportional to the confidence intervals.

4. The method of claim 1 , wherein finding the nearest neighbor includes using a k-nearest neighbor algorithm based on weighted distances to neighbors, the weighted distances to neighbors determined using a general pattern search algorithm.

5. The method of claim 3 , wherein the non-solar power generating device includes one or more of wind power device, a hydraulic power device, a coal power device, a nuclear power device, or a natural gas power device.

6. A method of analyzing solar power forecasts, the method comprising:
obtaining a test dataset of historical irradiance at a location of a solar power generating system;
normalizing the test dataset, the normalized dataset being independent of solar zenith angle;
clustering the test dataset into a plurality of weather classes, each weather class including a set of characteristics;
obtaining a forecast of irradiance for a given future time at the solar power generating system;
classifying the forecast into one of the plurality of weather classes;
determining confidence intervals of the forecast based on the set of characteristics of the one of the plurality of weather classes; and
based on the confidence intervals of the forecast, performing one of increasing output of a source of energy alternative to solar energy or decreasing output of the source of energy.

7. The method of claim 6 , wherein the set of characteristics includes an amount of cloud cover including at least one of a clear sky, an overcast sky, and a partly cloudy sky.

8. The method of claim 6 , wherein
the set of characteristics includes a degree of unpredictability of irradiance; and each of the plurality of weather classes has a different degree of unpredictability of irradiance.

9. The method of claim 6 , wherein the set of characteristics includes one or more of season, location, cloud formation, cloud class, or sky cover.

10. The method of claim 6 , wherein the forecast of irradiance is generated by a third party.

11. The method of claim 6 , further comprising:
generating electrical power using the solar power generating system; and
generating electrical power using the source of energy.

12. The method of claim 6 , wherein the source of energy includes one or more of wind power, hydraulic power, coal power, nuclear power, or natural gas power.

13. The method of claim 6 , wherein normalizing the test dataset includes dividing a point in the test dataset at a given time by a clear sky irradiance at the given time.

14. The method of claim 6 , wherein clustering the test dataset into a plurality of weather classes comprises:
plotting a histogram of the test dataset;
fitting a curve to the histogram, the curve including a combination of class-specific curves for each class;
extracting the class-specific curves for each of the classes; and
associating each point in the test dataset with one of the classes based on how closely each point fits to the class-specific curves.

15. The method of claim 6 , wherein
classifying the forecast into one of the plurality of weather classes comprises finding a nearest neighbor value of the forecast, the nearest neighbor value representing a point in the test dataset with similar weather conditions to the forecast; and
determining confidence intervals of the forecast is further based on a degree of unpredictability of irradiance of a class to which the nearest neighbor value belongs.

16. The method of claim 15 , wherein the nearest neighbor includes a different time of day than the forecast.

17. The method of claim 6 , wherein the solar power generating system comprises a plurality of individual power generating devices, including at least one solar power generating device, aggregated as a virtual power plant.

18. The method of claim 6 , further comprising modifying a price of the solar power based on the confidence intervals.

Assignees

• Fujitsu Ltd

Inventors

• He Dawei
• Alimohammadi Shahrouz

Classifications

• H02J3/004

  Generation forecast, e.g. methods or systems for forecasting future energy generation · CPC title

• H02J3/003

  Load forecast, e.g. methods or systems for forecasting future load demand · CPC title

• H02J2101/28

  Wind energy · CPC title

• H02J2101/24

  Photovoltaics · CPC title

• H02J2101/20

  Dispersed power generation using renewable energy sources · CPC title