Ensemble Machine Learning Model for Day Ahead Solar Power Forecasting for Mongolia Power System

The efficient and sustainable use of renewable energy sources requires an accurate power generation forecasting. This paper considers the application and comparison of ensemble algorithm based on decision tree for solar power forecasting for Mongolia power system. The input data for the machine learning model are historical data on power generation for 2019–2021 years of six photo-voltaic power plants operating in central power system of Mongolia and meteorological data in these power plants. Ensemble machine learning model allows to determine the non-linear and non-stationary dependence of the time series and besides can be implemented in the task of forecasting the daily generation schedule. The proposed model creates a day ahead forecast of hourly generation curve of the photo-voltaic power plants under consideration with a normalized absolute percentage error 6.5 – 8.4%. Increasing the accuracy of solar power forecasting can positively affect the operation and planning of the central power system of Mongolia.