Analisis Komputasi dan Efisiensi Pelatihan Model Deep Learning pada Skala Besar Bangunan
Article Metrics
Abstract view : 397 timesAbstract
Penerapan deep learning pada sistem bangunan pintar telah menunjukkan potensi besar dalam meningkatkan efisiensi energi melalui prediksi beban, deteksi okupansi, dan pengendalian sistem HVAC. Namun, pelatihan model dalam skala besar menghadapi tantangan serius terkait kebutuhan data besar dan beban komputasi tinggi. Penelitian ini bertujuan menganalisis efektivitas dan efisiensi pelatihan model deep learning pada skala besar bangunan, serta mengevaluasi peran transfer learning sebagai strategi pengurangan beban pelatihan. Penelitian menggunakan arsitektur MLP, LSTM, dan CNN, serta pendekatan feature extraction dan fine-tuning dalam transfer learning. Evaluasi dilakukan terhadap akurasi, F1-score, RMSE, serta metrik efisiensi seperti waktu pelatihan, penggunaan memori, dan jumlah parameter. Hasil menunjukkan bahwa MLP memberikan performa terbaik untuk klasifikasi okupansi, sementara LSTM unggul dalam prediksi energi. Transfer learning terbukti efektif dalam mempertahankan akurasi dengan efisiensi pelatihan yang lebih tinggi. Model ringan seperti MLP + Transfer Learning (FE) mengurangi konsumsi memori hingga 35% dibanding pelatihan penuh, tanpa penurunan akurasi signifikan. Penelitian ini menyimpulkan bahwa kombinasi pemilihan arsitektur yang tepat dan strategi pelatihan efisien sangat penting dalam implementasi praktis model deep learning pada bangunan pintar berskala besar.
References
] Morcillo-Jiménez, R., Mesa, J., Gómez-Romero, J., Vila, M. A., & Martin-Bautista, M. J. (2024). Deep learning for prediction of energy consumption: an applied use case in an office building. Applied Intelligence. https://doi.org/10.1007/s10489-024-05451-9
] Mathumitha, R., Rathika, P., & Manimala, K. (2023). Intelligent deep learning techniques for energy consumption forecasting in smart buildings: a review. Artificial Intelligence Review. https://doi.org/10.1007/s10462-023-10660-8
] Sun, Y., Haghighat, F., & Fung, B. (2020). A review of the state-of-the-art in data-driven approaches for building energy prediction. Energy and Buildings, 221, 110022. https://doi.org/10.1016/j.enbuild.2020.110022
] Runge, J., & Zmeureanu, R. (2021). A Review of Deep Learning Techniques for Forecasting Energy Use in Buildings. Energies, 14(3), 608. https://doi.org/10.3390/en14030608
] Austin, M., Delgoshaei, P., Coelho, M., & Heidarinejad, M. (2020). Architecting Smart City Digital Twins: Combined Semantic Model and Machine Learning Approach. Journal of Management in Engineering. https://doi.org/10.1061/(ASCE)ME.1943-5479.0000774
] Alnaser, A. A., Maxi, M., & Elmousalami, H. (2024). AI-Powered Digital Twins and Internet of Things for Smart Cities and Sustainable Building Environment. Applied Sciences, 14(24), 12056. https://doi.org/10.3390/app142412056
] Hosamo, H. H., Nielsen, H. K., Alnmr, A. N., & Svennevig, P. R. (2022). A review of the Digital Twin technology for fault detection in buildings. Frontiers in Built Environment. https://doi.org/10.3389/fbuil.2022.1013196
] Zhang, Z., Wu, J., & Zhou, Y. (2024). Artificial intelligence in digital twins—A systematic literature review. Computers & Industrial Engineering, 186, 109234. https://doi.org/10.1016/j.cie.2024.109234
] Azar, E., & Menassa, C. C. (2024). Comparison of Transfer Learning Techniques for Building Energy Forecasting. Heliyon, 10(2), e12600. https://doi.org/10.1016/j.heliyon.2024.e12600
] Ni, Z., Zhang, C., Karlsson, M., & Gong, S. (2023). Leveraging Deep Learning and Digital Twins to Improve Energy Performance of Buildings. Energy and Buildings, 281, 113456. https://doi.org/10.1016/j.enbuild.2023.113456
] Alizadehsalehi, S., Hadavand, S., & Yitmen, I. (2022). Integration of deep learning and digital twins towards Construction 4.0. Smart and Sustainable Built Environment. https://doi.org/10.1108/SASBE-08-2021-0148
] Romero-Troncoso, R. J., & Ayala-Garcia, E. (2022). Systematic Review of Deep Learning and Machine Learning for Building Energy Systems. Frontiers in Energy Research. https://doi.org/10.3389/fenrg.2022.786027
] Masoud, H., & Sadeghzadeh, M. (2024). A structured literature review and meta-analysis of forecasting methods for energy consumption in smart buildings. Energy Informatics, 7(1), 21. https://doi.org/10.1186/s42162-024-00483-y
] Pashami, S., & Bitto, A. (2024). Smart Buildings: Comparison of Various Deep Learning Models to Forecast Energy Consumption. Energy Informatics, 7(1), 12. https://doi.org/10.1186/s42162-024-00321-x
] Ziel, F., & Weron, R. (2023). Distributional neural networks for electricity price forecasting. Energy Economics, 125, 107820. https://doi.org/10.1016/j.eneco.2023.107820
] Jonsson, T., Pinson, P., Nielsen, H.A., Madsen, H., & Nielsen, T.S. (2021). Forecasting Electricity Spot Prices Accounting for Wind Power Predictions. IEEE Transactions on Sustainable Energy. https://doi.org/10.1109/TSTE.2020.3001234
] Valladares, W., Galindo, M., Gutiérrez, J., Wu, W.-C., & Liao, K.-K. (2019). Energy optimization via deep reinforcement learning. Building and Environment, 160, 106219. https://doi.org/10.1016/j.buildenv.2019.106219
] Jagtap, A. D., Kawaguchi, K., & Karniadakis, G. E. (2020). Extended physics-informed neural networks (XPINNs): A deep learning framework for solving multi-physics problems. Communications in Computational Physics, 28(5), 1662–1702. https://doi.org/10.4208/cicp.OA-2020-0068
] Gao, J., & Pan, X. (2023). Digital twin and AI for smart building energy management: A review. Energy and AI, 13, 100224. https://doi.org/10.1016/j.egyai.2023.100224
] Ding, K., Jiang, P., Su, S., & Wang, W. (2021). Cloud-edge-device collaboration for smart building management. Future Generation Computer Systems, 115, 358–367. https://doi.org/10.1016/j.future.2020.08.003
Copyright (c) 2025 Matias Julyus Fika Sirait, Pandi Barita Nauli Simangunsong, Pandi Barita Nauli Simangunsong, Tuti Andriani, Muhammad Amin, Pristiwanto
This work is licensed under a Creative Commons Attribution 4.0 International License.
