Method and system for estimating energy generation based on solar irradiance forecasting

Patent metadata

Field	Value
Publication number	US-11009536-B2
Application number	US-201616339240-A
Country	US
Kind code	B2
Filing date	Oct 5, 2016
Priority date	Oct 5, 2016
Publication date	May 18, 2021
Grant date	May 18, 2021

Abstract

Estimating energy generated by a solar system in a predetermined geographic area comprises, at each predetermined time instant: retrieving measured values of at least one weather parameter and of solar irradiance in the geographic area, the values related to a time slot before the predetermined time instant; performing auto-regression analysis of the measured values; estimating, based on the auto-regression analysis, a relationship between the at least one weather parameter and the solar irradiance; retrieving forecasted values of the at least one weather parameter in the geographic area, the forecasted values being forecasted for a second time slot after the predetermined time instant; performing regression analysis of the relationship between the at least one weather parameter and the solar irradiance of the forecasted values; forecasting solar irradiance in the second time slot based on the regression analysis, and estimating energy generated by the solar system in the second time slot.

First claim

The invention claimed is:

1. A method for estimating energy generated by a solar system in a predetermined geographic area, the method comprising, at each predetermined time instant:
retrieving measured values of at least one weather parameter and of solar irradiance in said predetermined geographic area, said measured values being related to a first time slot before said predetermined time instant;
performing an auto-regression analysis of the measured values;
estimating, based on said auto-regression analysis, a relationship between the at least one weather parameter and the solar irradiance, wherein said performing and said estimating comprise providing the measured values of the at least one weather parameter and of the solar irradiance to an encoder network of a sequence-to-sequence neural network, to obtain encoded measured values indicative of said relationship between the at least one weather parameter and the solar irradiance;
retrieving forecasted values of said at least one weather parameter in said predetermined geographic area, said forecasted values being forecasted for a second time slot after said predetermined time instant;
performing a regression analysis of said relationship between the at least one weather parameter and the solar irradiance of said forecasted values;
forecasting a solar irradiance in said second time slot based on said regression analysis wherein said forecasting comprises providing said encoded measured values and said forecasted values to a decoder network of the sequence-to-sequence neural network; and
estimating energy generated by the solar system in said second time slot according to the forecasted solar irradiance and at least one solar system parameter.

2. The method according to claim 1 , wherein said performing and said estimating comprise:
providing the measured values of at least one weather parameter and of solar irradiance to an encoder network of a sequence-to-sequence neural network, to obtain encoded measured values indicative of said relationship between the at least one weather parameter and the solar irradiance, and
providing the encoded measured values to an attention network to obtain a weighted combination of the encoded measured values, and
wherein said performing and said forecasting comprise:
providing said weighted combination of the encoded measured values and said forecasted values to a decoder network of the sequence-to-sequence neural network.

3. The method according to claim 1 , wherein:
the measured values comprise, for each weather parameter and for the solar irradiance, a plurality of measured values each one related to a respective measure time within said first time slot;
said retrieving comprises retrieving the measured values and the respective measure times; and
said providing the measured values of the at least one weather parameter and of solar irradiance to an encoder network comprises providing the measured values and the respective measure times to the encoder network.

4. The method according to claim 1 , wherein at least one of said at least one weather parameter is a categorical variable, each weather parameter other than a categorical variable being a continuous variable, and wherein said providing the measured values of the at least one weather parameter and of solar irradiance to an encoder network comprises:
mapping the measured values of each categorical variable to vectors of numbers in a continuous space,
concatenating the mapped measured values with the measured values of each continuous variable to obtain concatenated measured values of each weather parameter, and
providing said concatenated measured values to the encoder network together with the measured values of solar irradiance.

5. The method according to claim 4 , wherein said continuous space is a Euclidean space.

6. The method according to claim 1 , wherein said encoder network is a bi-directional recurrent neural network.

7. The method according to claim 1 , wherein said encoder network is a “Long Short Term Memory” or a “Gated Recurrent Unit” recurrent neural network.

8. The method according to claim 2 , wherein the attention network comprises a feed-forward network adapted to provide said weighted combination of the encoded measured values according to a current hidden state of the decoder network.

9. The method according to claim 1 , wherein the decoder network is a unidirectional decoder network.

10. The method according to claim 9 , wherein the decoder network is a “Long Short Term Memory” or a “Gated Recurrent Unit” recurrent neural network.

11. The method according to claim 1 , wherein said at least one weather parameter comprises at least one among:
textual general weather information;
temperature;
humidity; or
amount of sky occluded by clouds.

12. The method according to claim 1 , wherein said solar system is a photovoltaic system, and wherein the energy generated by the solar system is electric energy generated by the photovoltaic system.

13. A system for estimating energy generated by a solar system in a predetermined geographic area, the system comprising:
non-transitory computer readable media having computer-executable modules comprising:
a first module configured to retrieve, at a predetermined time instant, measured values of at least one weather parameter and of solar irradiance in said predetermined geographic area, said measured values being related to a first time slot before said predetermined time instant, and to retrieve forecasted values of said at least one weather parameter in said predetermined geographic area, said forecasted values being forecasted for a second time slot after said predetermined time instant;
a second module configured to estimate, based on an auto-regression analysis of the measured values, a relationship between the at least one weather parameter and the solar irradiance, wherein said second module comprises an encoder network of a sequence-to-sequence neural network, said encoder network providing encoded measured values indicative of said relationship between the at least one weather parameter and the solar irradiance;
a third module configured to forecast a solar irradiance in said second time slot based on a regression analysis of said relationship between the at least one weather parameter and the solar irradiance of said forecasted values, wherein said third module comprises a decoder network of the sequence-to-sequence neural network, the decoder network providing solar irradiance forecasting according to said encoded measured values and said forecasted values; and
a fourth module configured to estimate energy generated by the solar system in said second time slot according to the forecasted solar irradiance and at least one solar system parameter.

14. A system for estimating energy generated by a solar system in a predetermined geographic area, the system comprising:
a processor; and
non-transitory computer readable media having instructions stored thereon, wherein the instructions, when executed by the processor, cause the system to:
retrieve, at a predetermined time instant, measured values of at least one weather parameter and of solar irradiance in said predetermined geographic area, said measured values being related to a first time slot before said predetermined time instant;
retrieve forecasted values of said at least one weather parameter in said predetermined geographic area, said forecasted values being forecasted for a second time slot after said predetermined time instant;
estimate, based on an auto-regression analysis of the measured values, a relationship between the at least one wea

Assignees

• Telecom Italia Spa

Inventors

• Borean Claudio
• Grasso Ennio

Classifications

• G06N3/045

  Combinations of networks · 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

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• G06N3/09

  Supervised learning · CPC title