Алгоритми машинного навчання для прогнозування генерації СЕС

Abstract

The thesis discusses machine learning algorithms for forecasting the generation of solar power plants (SPPs), including linear models, support vector machines (SVM/SVR), ensemble methods (Random Forest, XGBoost) and neural networks (LSTM, CNN). Their advantages, limitations and optimal applications are identified, and a combined approach is proposed to improve the accuracy of forecasts depending on the type of data and time horizon.