Short-Term Energy Consumption Prediction in Iranian Buildings Using a Hybrid CNN-LSTM Model with Multimodal Data Fusion: A Case Study on Residential Buildings in Tehran

Niroumand, Mohammad; Jalili, Mohammad; Yarahmadi, Hossein

doi:10.61882/jgeri.3.2.23

Short-Term Energy Consumption Prediction in Iranian Buildings Using a Hybrid CNN-LSTM Model with Multimodal Data Fusion: A Case Study on Residential Buildings in Tehran

Document Type : Research article

Authors

Mohammad Niroumand ¹

Mohammad Jalili ¹

Hossein Yarahmadi ²

¹ Department of Architecture, Bo.C., Islamic Azad University, Borujerd, Iran

² Department of Computer Engineering, Bo.C., Islamic Azad University, Borujerd, Iran

10.61882/jgeri.3.2.23

Abstract

This study presents a hybrid CNN-LSTM model for short-term energy consumption prediction in Iranian residential buildings, focusing on Tehran. By integrating multimodal data, meteorological, temporal, occupancy proxies, and building metadata, and employing deep feature engineering via a stacked denoising autoencoder, the model achieves high accuracy (R² = 0.89) and robustness against data imperfections. The framework demonstrates the critical role of cultural and contextual features, such as Iranian holidays, in enhancing prediction validity. SHAP analysis provides interpretability, aligning model logic with local realities. The results offer a scalable, context-aware solution for intelligent energy management in Iran’s urban environment.‎

Highlights

❖ A hybrid CNN-LSTM model achieved superior accuracy (R²=0.89) for short-term energy forecasting in Tehran's residential buildings.
❖ The model uniquely integrated Iran-specific cultural calendar features, which were identified as top predictors of energy demand.
❖ A robust preprocessing pipeline using TimeGAN and VAE effectively handled the region's characteristic noisy and incomplete data.
❖ Multimodal fusion of meteorological, temporal, and occupancy proxy data captured the complex drivers of energy consumption.
❖ SHAP analysis provided critical model interpretability, validating its logic against local contextual patterns and realities.

Keywords

Short-term energy prediction

Hybrid CNN-LSTM

Deep feature engineering

Iranian residential buildings

Explainable AI

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. The ethical issues, including plagiarism, informed consent, misconduct, data fabrication and/or falsification, double publication and/or submission, redundancy, have been completely observed by the authors.

Credit Authorship Contribution Statement

Mohammad Niroumand: Conceptualization, Writing - original draft, Writing-review & editing. Mohammad Jalili: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Validation, Writing-review & editing. Hossein Yarahmadi: Formal analysis, Software, Writing-review & editing.

Bibliography

Mohammad Niroumand is a PhD candidate in Architecture at the Islamic Azad University, Borujerd Branch, and concurrently serves as an engineer with the Iran Construction Engineering Organization in Borujerd. His research is primarily centered on construction technology and sustainable building practices, with a focus on improving building performance. He has investigated the application of advanced materials, such as waterproof concrete incorporating crystalline and zycosilic technologies, for critical building execution levels. Niroumand has also explored the role of green roofs in reducing energy consumption, specifically within educational buildings, using a primary school for girls in Isfahan as a case study. His work in this area was published in both Persian and English in the Journal of Renewable and New Energy in 2021. With publications appearing in journals like Építőanyag, his academic work bridges material science and sustainable architectural design. His dual role as a PhD researcher and a practicing engineer allows him to address practical challenges in the construction industry from a research-informed perspective.

Mohammad Jalili is an academic and researcher with a background as the former Head of the Faculty of Construction Industry and Environment at the Islamic Azad University, Borujerd Branch. His research focuses on the intersection of urban design, environmental psychology, and community perception. He has explored the relationship between urban spatial configurations and spirituality, notably in a 2025 study on historic mosques in Isfahan's Naqsh-e Jahan Square using Space Syntax methodology. Jalili has also investigated fear of crime and social cohesion in different community types within Tehran. His work extends to the educational environment, examining the impact of school physical quality on attention deficit hyperactivity disorder in children. His recent publications appear in journals such as the International Journal of Urban Sciences and the Culture of Islamic Architecture and Urbanism Journal. His scholarly contributions are documented on his Google Scholar profile, which lists his work from 2025. Through his diverse research, Jalili addresses critical aspects of how the built environment influences human behavior and well-being.

Hossein Yarahmadi is an assistant professor in the Department of Computer Engineering at the University of IAU Borujerd Branch, Iran. He received his Ph.D. in Computer Science (Software) from the University of Antwerp (Belgium) and his Ph.D. in Computer Engineering (Software Systems) from the IAU Science and Research Branch (Iran) in collaboration in 2023. He also holds an M.S. degree in computer software engineering from the IAU Arak branch (2004) and a B.S. degree in computer software engineering from the University of Isfahan (Iran) in 2001. From September 2001 to September 2023, he served as a lecturer at the University of IAU Borujerd Branch, Iran. During this time, he collaborated with the University of ABRU, Iran, from 2013 to 2023. For one year, from Oct 2023 to Oct 2024, He was a postdoctoral researcher fellow at the University of Antwerp (Universiteit Antwerpen), Belgium, working in the MICCS lab as a sub-lab of AnSyMo. His research interests include multi-agent systems, game theory, machine learning, and information security.

Citation
M. Niroumand, M. Jalili, and H. Yarahmadi, "Short-Term Energy Consumption Prediction in Iranian Buildings Using a Hybrid CNN-LSTM Model with Multimodal Data Fusion: A Case Study on Residential Buildings in Tehran," Journal of Green Energy Research and Innovation, vol. 3, no. 2, pp. 23-35, 2026.

[1] D. Liu, "International Energy Agency (IEA)," The Palgrave Encyclopedia of Global Security Studies, pp. 830–836, 2023.

[2] C. Fan, F. Xiao, and J. Wang, "A short-term building cooling load prediction method using deep learning architectures," Applied Energy, vol. 195, pp. 222–233, 2017.

[3] D. B. Crawley, J. W. Hand, M. Kummert, and B. T. Griffith, "A new method for comparing energy simulation programs," in Proc. ASHRAE/DOE/BTECC Conf., 2001, pp. 23–26.

[4] K. Amasyali, and N. M. El-Gohary, "A Review of Data-Driven Building Energy Consumption Prediction Studies," Renewable and Sustainable Energy Reviews, vol. 81, pp. 1192–1205, 2018.

[5] I. Goodfellow, J. Pouget-Abadie, et al., "Generative Adversarial Networks," Communications of the ACM, vol. 63, no. 11, pp. 139–144, 2020.

[6] S. Hochreiter and J. Schmidhuber, "Long short-term memory," Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997.

[7] L. Qiu, "Deep learning approaches for building energy consumption prediction: A systematic review," Energy and Sustainable Energy, vol. 2, no. 3, pp. 11–17, 2024.

[8] N. Somu, M. R. Raman, and K. Ramamritham, "A deep learning framework for building energy consumption forecast," Renewable and Sustainable Energy Reviews, vol. 137, p. 110591, 2021.

[9] H. Jang and J. Kang, "A stochastic model of integrating occupant behaviour into energy simulation with respect to actual energy consumption in high-rise apartment buildings," Energy and Buildings, vol. 121, pp. 205–216, 2016.

[10] D. Lahat, T. Adali, and C. Jutten, "Multimodal data fusion: An overview of methods, challenges, and prospects," Proc. IEEE, vol. 103, no. 9, pp. 1449–1477, 2015.

[11] I. Goodfellow, J. Pouget-Abadie, et al., "Generative Adversarial Networks," Communications of the ACM, vol. 63, no. 11, pp. 139–144, 2020.

[12] M. M. Forootan, I. Larki, R. Zahedi, and A. Ahmadi, "Machine Learning and Deep Learning in Energy Systems: A Review," Sustainability, vol. 14, no. 8, 4832, 2022.

[13] M. A. Mirjalili, A. Aslani, R. Zahedi, and M. Soleimani, "A Comparative Study of Machine Learning and Deep Learning Methods for Energy Balance Prediction in a Hybrid Building-Renewable Energy System," Sustainable Energy Research, vol. 10, no. 1, 2023.

[14] P. P. Hamedany, "A Survey on Ambient Intelligence Contexts: A Context-Aware Taxonomy Based on Deep Learning and Internet of Things Synergy," IEEE Access, vol. 12, pp. 12345–12367, 2024.

[15] D. B. Unsal, A. Aksoz, S. Oyucu, J. M. Guerrero, and M. Guler, "A Comparative Study of AI Methods on Renewable Energy Prediction for Smart Grids: Case of Turkey," Sustainability, vol. 16, no. 7, 2894, 2024.

[16] J. Yoon, J. Jordon, and M. van der Schaar, "Time-series generative adversarial networks," in Advances in Neural Information Processing Systems, vol. 32, pp. 5622–5631, 2019.

[17] C. Fan, J. Y. Wang, W. J. Gang, and S. H. Li, "Assessment of deep recurrent neural network-based strategies for short-term building energy predictions," Applied Energy, vol. 236, pp. 700–710, 2019.

[18] W. J. Clancey, "The conceptual nature of knowledge, situations, and activity," in Human and Machine Expertise in Context, vol. 247, pp. 291, 1997. [TITLE MISMATCH]

[19] K. Singh Sian, "Frugal Innovation – “Doing More with Less”," LAND.TECHNIK 2022, pp. 193–202, 2022.