Occupant-Aware Deep Learning Model for Energy Prediction in Smart Buildings

Building energy prediction plays a crucial role in reducing energy wastage and ensuring the storage of sufficient sustainable energy for future use. However, existing building energy prediction models exhibit low accuracy, and their applicability is typically limited to specific use cases. Therefore, this paper proposes a novel hybrid model that combines Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) for building energy prediction. The proposed hybrid model captures spatial patterns and features in the data using CNN. Furthermore, it captures temporal dependencies and long-term relationships by using BiLSTM. This paper improves energy prediction accuracy by integrating occupancy count as an additional input into the proposed hybrid model. This also improves the practical applicability of the proposed energy prediction model. The performance of the proposed hybrid model is extensively and rigorously validated against various evaluation metrics, including prediction accuracy, Mean Absolute Error (MAE), and Mean Squared Error (MSE). The simulation results show that the proposed model outperforms existing state-of-the-art approaches. The model provides improved accuracy and enables stakeholders to optimize energy usage and enhance the overall energy efficiency in buildings.