Estimating Solar Power Plant Data Using Time Series Analysis Methods

Abstract

When meteorological data such as temperature, precipitation, weather events and economic data such as stock prices and exchange rates reach large levels, it may be necessary to analyze them with time series analysis methods. The aim of this research is to analyze the data of solar power plants with time series and make predictions for the future. To achieve this goal, solar panel data with historical depth will be collected, the collected data will be trained and predicted by various time series analysis methods and comparison will be made according to the prediction success among the related models. Methodology: With this study, using Python 3.6 and R 3.6.1, the time series estimation models were modeled with AR, ARMA, SARIMA, DES and TES, the difference between the real value and the predicted value of the data was found by the RMSE (Square Root of the Mean Square Error) method and it was seen which model has the best ability to estimate the dataset. In addition, with the trend and seasonality of the data, detailed information about the dataset was obtained with descriptive analysis and graphics. As a result, it was seen that using SARIMA or TES models in the datasets that show seasonal change in the light of the studies and estimations performed gives better results.