Load forecasting of the power system: An investigation based on the method of random forest regression

Abstract

Accurate power load forecasting plays an important role in the power dispatching and security of grid. In this paper, a mathematical model for power load forecasting based on the random forest regression (RFR) was estab-lished. The input parameters of RFR model were determined by means of the grid search algorithm. The prediction results for this model were compared with those for several other common machine learning methods. It was found that the coefficient of determination (R2) of test set based on the RFR model was the highest, reaching 0.514 while the corresponding mean absolute error (MAE) and the mean squared error (MSE) were the lowest. Besides, the impacts of the air conditioning system used in summer on the power load were discussed. The calculation results showed that the introduction of indexes in the field of Heating, Ventilation and Air Conditioning (HVAC) could improve the prediction accuracy of test set.