A Hybrid Model for the Prediction Of Electrical Energy Consumption Using Hybrid LSTM and ML Regressors

Accurate forecasting of energy consumption over long periods is extremely important for companies that distribute and supply electricity, whether from the government or private sector. It is necessary in terms of improving the quality of energy production in the future, especially in countries like Iraq that have been suffering from an energy crisis for a long time. This study used electricity consumption data from the Ministry of Electricity in Iraq for the city of Baghdad, specifically the Rusafa area, for the years from 2021 to 2023. In this study, several models were worked on and compared with the proposed hybrid model (CNN-Stacked Bi-LSTM) with RF and KNN to achieve better performance in classification or prediction tasks. To predict future electricity consumption and improve the quality of energy production, the models were trained on electrical energy consumption data. We trained the models on (30) epochs, taking the MAPE and RMSE resulting from our assessment of the quality of energy consumption. The experiments found that the best results is the hybrid model using RF regressor, which produced a result of MAPE: 0.195046, RMSE: 0.101919 and MAE: 0.078101.