An intelligent deep learning-based approach for electrical energy consumption forecasting in buildings using smart meter data

Abstract— Forecasting electricity demand is a great way to plan ahead for improving the power generator sector and getting ready for regular operations. The rapid growth of cities has led to a sharp increase in energy demand, resulting in a significant discrepancy between the quantity of energy produced and consumer demand. In order to close the gap between supply and demand, a new prediction model is needed that uses automated methods to dynamically forecast the building’s energy consumption. In the proposed method, an intelligent deep learning technique LSTM (Long short-term memory) is accompanied by pre-processing for the forecasting power consumption in a building. The proposed model is evaluated over benchmark datasets that demonstrate a significant reduction in the error rate compared to other methods. The proposed LSTM method achieves better prediction results, thus demonstrating its effectiveness with other methods. Root mean square error (RMSE), mean square error (MSE), and mean absolute error (MAE) are used to evaluate the proposed model. The results indicated that the proposed model error vanishes to reach an RMSE value of 1%, MSE value of 0.3%, and MAE value of 1%. RNN (Recurrent Neural Network), GRU (Gated Recurrent Unit), and Bi-LSTM (Bi-directional Long Short-Term Memory) are taken into consideration for the comparative study, and the research is done to predict energy consumption one month in advance, with 80% of the data being used as the training set and the remaining 20% being used for testing.