Due to the increasing consumption in energy, accurate building energy prediction has become important, and recent studies achieve significant advancement with deep learning-based approaches. However, there are still some challenges. For accurate prediction, a complex structure that considers both temporal and spatial features must be applied, which makes it difficult to extend to other cases. Furthermore, some recent studies use simple time-series model as comparison, which may undermine the advantage of incorporating spatial features. In this paper, we propose novel Two-stage Spatiotemporal Prediction Module (TSPM) to overcome these limitations. Our TSPM can incorporate spatial features based on building floor plans images which can easily obtained. In addition, we conduct fair comparisons with promising time-series models to demonstrate effectiveness of our approach.
Keywords
Acknowledgement
This work was supported by Institute of Information & communications Technology Planning & Evaluation (IITP) under the Artificial Intelligence Convergence Innovation Human Resources Development (IITP-2023-RS-2023-00256629, 50%) grant funded by the Korea government(MSIT). This work was supported by the Institute of Information & Communications Technology Planning & Evaluation(IITP)-Innovative Human Resource Development for Local Intellectualization program grant funded by the Korea government(MSIT)(IITP-2025-RS-2022-00156287, 50%).