Short-term Solar Power Generation Forecasting for Microgrid

Abstract

Nowadays, the world's energy consumption is growing, and solving the problem of replacing traditional sources with alternative ones is urgent. The solution to this problem is impossible without prior data analysis and further forecasting of energy production from alternative sources. However, the use of alternative energy sources in the wholesale electricity and power market is impossible without the implementation of short-term forecasting models for the day ahead. In this article, the authors briefly analyze the existing methods of short-term forecasting, which are used in forecasting power generation by solar power plants. Currently, there are already a large number of short-term forecasting models, and they all differ in their characteristics. Therefore, in order to highlight the most promising for further use and short-term forecasting models development, the authors analyzed existing forecasting methods in the energy domain. During the study, the primary processing of the initial data obtained from existing solar power plants was performed using spectral analysis. Next, to build a prognostic model, a correlation analysis of the primary data was performed, which showed the absence of a linear relationship between the retrospective data components. The authors, based on the indications of correlation analysis, decided to select parameters for the construction of a prognostic model empirically. As a result of the study, a mathematical model based on an artificial neural network was proposed and a training sample was formed. The architecture of the artificial neural network was determined, the result of which is a short-term power generation forecast from solar panel.