Energy management and load forecasting in Smart grid

Energy consumption has increased considerably over the past few decades. This expansion then suggests a future interest in the use of electricity wholesalers. Predicting energy consumption requires in this work to propose two approaches, one based on long short-term memory (LSTM) and the other based on simple RNNs based on recurrent neural networks. These models take into account electricity consumption. The modeling data are from Global Energy Forecasting 2012. To evaluate the performance of the two models, four techniques are used: MSE (Mean Squared Error), MAE (Mean Absolute Error), RMSE (Root Mean Square Error) and MAPE (Mean Absolute Percentage Error) in order to validate the best available method