Reduction of the Risk of Inaccurate Prediction of Electricity Generation from PV Farms Using Machine Learning

Problems with inaccurate prediction of electricity generation from photovoltaic (PV) farms cause severe operational, technical, and financial risks, which seriously affect both their owners grid operators. Proper results are required for optimal planning the spinning reserve as well managing inertia frequency response in case contingency events. In this work, impact a number meteorological parameters on PV Poland was analyzed using Pearson coefficient. Furthermore, seven machine learning models Lasso Regression, K–Nearest Neighbours Support Vector AdaBoosted Regression Tree, Gradient Boosted Random Forest Artificial Neural Network were developed to predict 0.7 MW solar power plant Poland. The evaluated determination coefficient (R2), mean absolute error (MAE), root square (RMSE). It found out that horizontal global irradiation water saturation deficit have strong proportional relationship systems. All proposed turned perform predicting farm. most reliable accurate model, it received highest R2 (0.94) lowest MAE (15.12 kWh) RMSE (34.59 kWh).