Day-ahead load forecasting using ensemble empirical mode decomposition with random forest model considering the effect of solar irradiance and heat index

Abstract This research proposes a hybrid approach for short-term load forecasting (STLF) using Random Forest (RF) model with an Ensemble Empirical Mode Decomposition (EEMD) based time-frequency data analysis method to enhance load prediction in electric power networks. The study is carried out in two steps: (a) first, the EEMD method is applied to six-month hourly load data from 1st January 2024 to 30th June 2024 (i.e., 181 days for 24 h) of Bengaluru city (Karnataka, India). In the second step, Random Forest (RF) method is applied to obtain the aggregated final forecasted output for the winter period, spring period, and summer period. The novelty of this research work is that, along with meteorological variables (MV), i.e., relative humidity, dew point, dry bulb temperature, the effect of solar irradiance (SR) and heat index (HI), is also taken into account for precise load forecasting. The efficacy of the proposed EEMD-RF model is validated and analysed considering various multi-meteorological variables and HI by comparing with various single and hybrid STLF methods. The statistical and comparative study shows that the mean absolute percentage error (MAPE) of the proposed hybrid model is 1.08%.