Energy Consumption Prediction in Electrical Appliances of Commercial Buildings Using LSTM-GRU Model

Abstract

As with economic growth and urbanization, there is a significant impact on energy consumption in residential and commercial buildings. Analyzing the energy consumption of buildings is not a simple task to perform so it's much necessary to design an effective building energy management system, which can be helpful to evaluate the energy efficiency of different building structures. Recently, artificial intelligence, machine learning, and deep learning models have become most useful in the field of prediction and forecasting. This research presents a unique deep learning model using LSTM and GRU recurrent neural network (RNN) to predict the exact pattern of time series data for predicting building appliances energy consumption. The model is trained for the required features and evaluated by comparing the actual and predicted values. We have performed the analysis using a benchmark appliance energy data set and have taken metrics such as error rate, loss value, mean squared error (MSE), mean absolute error (MAE), mean absolute percentage error (MAPE) value, prediction accuracy, and model accuracy to evaluate the performance of the model. The outcome of this work shows that GRU exhibits better performance and achieved the minimum value of root mean square error and model loss.