Predicting Energy Consumption Using LSTM and CNN Deep Learning Algorithm

Abstract

Innovations in technologies that rely on electricity have led to an uncontrollable rise in power usage. In order to predict future electricity demand and enhance the power distribution system, analysis and forecasting of energy consumption systems are necessary. Several issues with the present energy consumption prediction methods make it difficult to anticipate actual energy usage with any degree of accuracy. In order to master the energy prediction method, this study examines fourteen years' worth of hourly energy usage data from a Kaggle open source dataset. In addition, a Long Short Term Memory (LSTM) and Convolution Neural Network (CNN) based method for estimating energy consumption based on actual datasets is presented in the research. The empirical findings demonstrate which LSTM and CNN architectures can improve energy consumption forecasting accuracy.