KR-20260067180-A - ELECTRONIC DEVICE AND OPERATING METHOD FOR PREDICTING SOLAR POWER GENERATION IN A SPECIFIC AREA USING A PREDICTION MODEL BASED ON DELAUNAY TRIANGULATION

KR20260067180AKR 20260067180 AKR20260067180 AKR 20260067180AKR-20260067180-A

Abstract

According to various embodiments, an electronic device for predicting solar power generation in a specific region using a Delaunay triangulation-based prediction model includes a processor, said processor receives location information data of said specific region located in a first region to which the Delaunay triangulation is applied, and is configured to predict solar power generation in said specific region by applying the location information data of said specific region to the prediction model as input data of said prediction model, said prediction model may be learned based on location information data of each vertex of said Delaunay triangulation generated based on location information data of all weather stations located within said first region, weather data of each of said weather stations located within said first region, and the daily power generation time of said solar power plants based on location information data of said solar power plants located within said first region. According to various embodiments, a method of operation of an electronic device for predicting solar power generation of a specific region using a Delaunay triangulation-based prediction model comprises: receiving location information data of the specific region located in a first region to which the Delaunay triangulation is applied; and applying the location information data of the specific region to the prediction model as input data of the prediction model to predict the solar power generation of the specific region; wherein the prediction model may be learned based on the daily power generation time of all solar power plants according to the location information data of each vertex of the Delaunay triangulation generated based on the location information data of all weather stations located within the first region, the weather data of each weather station located within the first region, and the location information data of all solar power plants located within the first region. Various other embodiments are also possible.

Inventors

신종호
한용수
소민섭

Assignees

조선대학교산학협력단

Dates

Publication Date: 20260512
Application Date: 20241105

Claims (5)

In an electronic device for predicting solar power generation in a specific region using a Delaunay triangulation-based prediction model, Includes a processor, The above processor is, Receiving location information data of the specific area located in the first area to which the above Delaunay triangulation is applied, and As input data for the above prediction model, location information data of the above specific region is applied to the prediction model to predict the amount of solar power generation of the above specific region, and is configured to predict the amount of solar power generation of the above specific region. The above prediction model is learned based on the location information data of each of the Delaunay triangulation vertices generated based on the location information data of all weather stations located within the first area, the weather data of each of all weather stations located within the first area, and the daily power generation time of all solar power plants based on the location information data of all solar power plants located within the first area. Electronic device.
In Article 1, The above location information data is, including coordinate data, latitude data, and longitude data corresponding to the above location, Electronic device.
In Paragraph 2, The above weather data is, Including a specific number of pre-set variables selected by applying PFI (Permutation feature importance) to all factors affecting the power generation of all weather stations located within the first area, Electronic device.
In Paragraph 3, The above prediction model is, Find the three vertices of the Delaunay triangulation triangle corresponding to the location information data of the first solar power plant, which is one of all solar power plants located within the first area, and By applying the location data of the three weather stations corresponding to the three vertices and the location data of the first solar power plant to the Haversine formula, the first distance, the second distance, and the third distance, which are the distances between the first solar power plant and each of the three vertices, are calculated, and By applying the first distance, the second distance, and the third distance to the following [Mathematical Formula 1], the weights for the first distance, the second distance, and the third distance are calculated, respectively, and Learning based on the daily power generation time of the solar power plant calculated by applying weights for the first distance, the second distance, and the third distance to the weather data of the solar power plant corresponding to the first distance, the second distance, and the third distance, Electronic device. [Mathematical Formula 1] Here represents the weight for the i-th distance, and means the i-th distance.
A method of operation of an electronic device for predicting solar power generation in a specific region using a Delaunay triangulation-based prediction model, A step of receiving location information data of the specific region located in the first region to which the above Delaunay triangulation is applied; and A step of predicting the solar power generation amount of the specific region by applying location information data of the specific region to the prediction model as input data for the prediction model; Includes, The above prediction model is learned based on the location information data of each of the Delaunay triangulation vertices generated based on the location information data of all weather stations located within the first area, the weather data of each of all weather stations located within the first area, and the daily power generation time of all solar power plants based on the location information data of all solar power plants located within the first area. Method of operation of an electronic device.

Description

Electronic device and operating method for predicting solar power generation in a specific area using a prediction model based on Delaunay triangulation The present invention relates to the prediction of solar power generation and the selection of optimal power plant locations. More specifically, it relates to a model and system capable of predicting the power generation of a solar power plant by augmenting data from weather stations using the Delaunay triangulation algorithm. Furthermore, it provides a technology that improves prediction accuracy by ensuring spatial and temporal consistency of weather variables, and enables more precise analysis of the optimal location selection and investment feasibility of a solar power plant. With the rapid surge in interest and demand for renewable energy sources, particularly solar power, as a means to reduce carbon emissions and minimize environmental impact in response to climate change and increasing global energy demand, the need for power generation forecasting models with high prediction accuracy is emerging to maximize the operational efficiency of solar power plants as the utilization of renewable energy expands. Technology for accurately predicting solar power generation is not only essential for stable power supply and meeting demand, but also plays a crucial role in cost reduction, efficient power management, power plant site selection, and investment feasibility assessment. Most existing solar power generation prediction models rely on a method that simply augments data collected from weather stations with adjacent data. While this ensures the representativeness of power plant locations and weather data, this approach fails to properly reflect the spatial distance and relationship between power plants and weather stations, often resulting in reduced prediction accuracy or causing significant errors under specific weather conditions. This is a problem arising from the failure to reflect the spatial relationships between data stations. Particularly in areas where weather stations are not uniformly distributed, weather data from specific stations is not suitable for power plant locations; consequently, existing methods suffer from insufficient accuracy of weather data for power plant locations, leading to limitations in prediction performance. To overcome these limitations, various techniques to spatially augment meteorological data and enhance consistency are currently being researched. Among these, the Delaunay triangulation algorithm is attracting attention as an effective method that more accurately reflects the spatial relationships between observation stations and can precisely augment meteorological data. Delaunay triangulation is a technique that forms triangles by connecting each weather station, such that the three sides of the triangle represent the shortest distances between adjacent stations. This algorithm ensures spatial connectivity, allowing station data to be combined more consistently with power plant locations, and can contribute to improving prediction accuracy. In addition, a process of correcting data from adjacent stations is necessary to compensate for missing or omitted meteorological data; in particular, in solar power generation forecasting, the distance between the power plant location and the adjacent weather station acts as a significant variable affecting prediction performance. Accordingly, research is being conducted on a method to augment weather data by utilizing the Haversine formula to calculate the distance between power plants and each weather station and assigning distance-based weights. Since the Haversine formula calculates the distance between two points by considering the curvature of the Earth, it can accurately reflect the physical distance between the power plant location and adjacent observation stations. Through this weight-based reinforcement technique, data from the weather station closest to the power plant is prioritized, thereby improving the reliability of the prediction model. Finally, attempts are being made to improve prediction accuracy by diversifying the types and quality of meteorological data, and various meteorological variables such as solar radiation, sunshine duration, temperature, and humidity are used as important input data in prediction models because they have a close influence on solar power generation. Existing solar power generation prediction models often consider single variables such as solar irradiance or temperature; however, this approach has limitations in accurate prediction because it fails to reflect the interactions between multiple meteorological variables. To address this, there is a need for a technology in this invention that improves the accuracy of power generation prediction by combining various meteorological variables with the Delaunay triangulation technique. FIG. 1 illustrates a block diagram of an electronic device according to various embodiments of the present