KR-102961967-B1 - Method and System for Predicting Renewable Energy Power Generation using Data from Multiple Weather Stations

KR102961967B1KR 102961967 B1KR102961967 B1KR 102961967B1KR-102961967-B1

Abstract

The method for predicting renewable energy generation using multiple weather station data according to the present invention may include: a step of dividing a target planar area into a plurality of triangular regions with weather stations as vertices; a step of clustering renewable energy power plants in each divided triangular region; a step of identifying weather stations located at the vertices of the triangular regions where the power plants are clustered in order to predict the generation amount of each renewable energy power plant; and a step of predicting the generation amount of the power plant by applying weather information of the identified weather stations to a prediction model.

Inventors

문종희
권성철
김홍석
정재익
송근주

Assignees

한국전력공사

Dates

Publication Date: 20260508
Application Date: 20221122

Claims (12)

A step of dividing the target planar region into multiple triangular regions with weather stations as vertices; A step of clustering renewable energy power plants in each divided triangular region; To predict the power generation amount of each new and renewable energy power plant, the step of identifying weather stations located at the vertices of a triangular region in which the power plants are clustered; and A step of predicting the power generation of the power plants clustered into the corresponding areas for each triangular area by applying weather information from the weather stations identified in the prediction model. Includes, The step of predicting the power generation of the above-mentioned power plant is, A method for predicting renewable energy generation using multiple weather station data, wherein solar power plants distributed in a divided area are used to predict solar power generation by inputting weather information from three weather stations corresponding to vertices during prediction.
In paragraph 1, The step of predicting the power generation of the above-mentioned power plant is, A step of collecting weather information from identified weather stations and applying it to a prediction model for power generation prediction; and The step of predicting the future power generation of the above power plant in the above prediction model A method for predicting renewable energy generation using data from multiple weather stations including
In paragraph 1, Step of determining the entire target planar area for performing the renewable energy generation prediction method A method for predicting renewable energy generation using data from multiple weather stations, including additional
In paragraph 1, Step of training the prediction model using weather information from the identified weather stations A method for predicting renewable energy generation using data from multiple weather stations, including additional
In paragraph 1, In the step of dividing into the above multiple triangular regions, A method for predicting renewable energy generation using data from multiple weather stations that performs region partitioning using the Delaunay triangulation technique.
In paragraph 1, The above prediction model is, A method for predicting renewable energy generation using data from multiple weather stations, which is a TransGRU model in which a Transformer encoder and a Gate Circulation Unit (GRU) are fused.
A regional clustering unit that divides a target planar area into multiple triangular regions with weather stations as vertices, and clusters new and renewable energy power plants in each divided triangular region; A weather data collection unit that identifies weather stations located at the vertices of a triangular region where the power plants are clustered, in order to predict the power generation amount of each new and renewable energy power plant, and collects weather information from the identified weather stations; and A prediction model that predicts the future power generation of solar power plants distributed in a segmented area by inputting weather information from three weather stations corresponding to the vertices during prediction and utilizing it for solar power generation forecasting. A renewable energy generation prediction system using data from multiple weather stations including
In Paragraph 7, A storage unit that records information regarding the division of the above-mentioned target planar area into multiple triangular regions with weather stations as vertices, and information regarding new and renewable energy power plants assigned (clustered) to each triangular region. A renewable energy generation prediction system using data from multiple weather stations that further includes
In Paragraph 7, A learning unit that trains the prediction model using weather information from the identified weather stations accumulated over a past period and operation information of the renewable energy power plant. A renewable energy generation prediction system using data from multiple weather stations that further includes
In Paragraph 7, A power plant operation information collection unit that collects operation information of the renewable energy power plant from a power generation control device or an upper server of the renewable energy power plant. A renewable energy generation prediction system using data from multiple weather stations that further includes
In Paragraph 7, The above-mentioned regional clustering unit is, A renewable energy generation prediction system using data from multiple weather stations that performs region partitioning using the Delaunay triangulation technique.
In Paragraph 7, The above prediction model is, A renewable energy generation prediction system using data from multiple weather stations, which is a TransGRU model in which a Transformer encoder and a Gate Circulation Unit (GRU) are fused.

Description

Method and System for Predicting Renewable Energy Power Generation using Data from Multiple Weather Stations The present invention relates to a method for easily predicting solar power generation for newly introduced solar power plants by applying the Delaunay technique to improve prediction performance through a prediction model that uses multiple weather stations as input conditions, and applying Delaunay triangulation based on weather stations. Globally, various policies are being implemented to reduce greenhouse gas emissions as part of efforts to address climate change following the Paris Agreement. In Korea, the expansion of new and renewable energy has been presented as the most effective reduction measure, leading to a rapid increase in renewable power sources within the power grid, with solar power sources driving the adoption of domestic renewable energy. Since solar power sources are dependent on weather conditions, they have intermittent and uncertain output characteristics, which can degrade the stability of the power grid. To address the issues with solar power sources, the government introduced a small-scale power brokerage business that aggregates small-scale power resources to sell electricity on behalf of power generators, as well as a renewable energy generation forecasting system that requires renewable energy generation volumes to be predicted and submitted one day in advance, with settlement payments provided if the volume is met within a certain margin of error on the day. Furthermore, accurate forecasting of solar power generation is required to resolve the uncertainty of such solar power generation and ensure a stable power supply, and many studies are being conducted to reduce forecasting errors. Prediction research has evolved from statistical methods to the recent emergence of AI-based prediction techniques. A representative example of deep learning is a Long Short-Term Memory (LSTM)-based prediction model that considers the time-series characteristics of solar power generation data according to weather conditions. Based on these prediction technologies, various weather elements from weather stations are used as input data to design a solar power generation prediction model that meets the conditions. Subsequently, the model is trained to output results, which are then verified for validity by comparing them with actual solar power generation. In this case, the single weather station closest to the solar power source is selected to acquire meteorological data. However, since solar power generation forecasting is influenced by geographical characteristics and weather conditions, even when using data from the closest weather station, there are limitations in improving the prediction accuracy of the forecasting model due to meteorological errors caused by the physical distance between the solar power plant and the weather station. In conclusion, to improve the accuracy of solar power generation forecasts, it is necessary to reduce the error caused by the physical distance between solar power plants and weather stations. FIG. 1 is a flowchart illustrating an embodiment of a method for predicting renewable energy generation using data from multiple weather stations according to the concept of the present invention. FIG. 2 is a structural diagram of a solar photovoltaic prediction system capable of performing a method for predicting renewable energy generation using data from multiple weather stations of FIG. 1. Figures 3a and 3b are conceptual diagrams illustrating the Delaunay triangulation technique. Figures 3c and 3d are diagrams illustrating Delaunay triangulation and photovoltaic power source clustering based on domestic weather stations. FIG. 4 is a structural diagram illustrating a Transformer encoder of a multi-weather station data-based prediction model that can be applied to the present invention. Figure 5 is a structural diagram illustrating the GRU model. Figure 6a is an example of applying Voronoi partitioning based on domestic weather stations, and Figure 6b is an example of applying photovoltaic power source clustering to the partitioned regions. Figures 7a and 7b are graphs showing the prediction results for the Bornoi and Delaunay triangulation methods. FIG. 8 is a block diagram illustrating an embodiment of a renewable energy generation prediction system using data from multiple weather stations according to the concept of the present invention. In describing the present invention, terms such as first, second, etc. may be used to describe various components, but the components may not be limited by the terms. The terms are intended solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component. When it is mentioned that a component is connected to or coupled with anothe