Multi-Horizon Energy Forecasting Using a Hybrid CNN-BiLSTM LSTM Deep Learning Framework

Abstract

Energy forecasting plays an important role in energy management. The fluctuations in the consumption of energy at different time intervals lead to the difficulty in making accurate predictions about future energy usage. In Multi-Horizon Energy Forecasting Using a Hybrid CNN-BiLSTM LSTM Deep Learning Framework, deep learning methods have been used to identify spatiotemporal characteristics within the data for forecasting purposes. The combination of CNN, Bi-LSTM, and LSTM helps the model learn about the complexities and inter relationships in the data better.The use of XAI increases the transparency of the process. The performance of the model has been evaluated based on the Individual Household Electric Power Consumption dataset which consists of over two million records. The mean squared error obtained was 0.01605 whereas the mean absolute error was 0.010389. This framework outperforms others in terms of predictive power making the study highly relevant and superior to other approaches.