本研究針對空調箱（Air Handling Unit, AHU）系統的故障診斷問題，提出了一種基於非監督式學習的領域適應方法，旨在解決不同類型空調箱數據之分佈差異導致的模型性能下降問題。研究背景基於能源效率提升的需求以及暖通空調系統在辦公場所中的耗電占比高達58.62%的現狀。針對雙管道空調箱（Dual Duct Air Handling Unit, DDAHU）與單管道空調箱（Single Duct Air Handling Unit, SDAHU）的數據分佈不一致問題，本研究提出利用雙管道系統作為源域（source domain），單管道系統作為目標域（target domain），進行無標籤的跨領域錯誤診斷。
本方法以領域適應（Domain Adaptation, DA）為核心，採用時間與頻率特徵編碼器來提取數據特徵，通過統計差異對齊（如Sinkhorn Divergence與最大均值差異MMD）減少兩域之間的分佈偏移。在特徵修正階段，輔助解碼器被用於重構原始數據，從而優化特徵表示並提高模型在無標籤數據上的識別準確性。本研究同時結合了特徵重要性分析技術，通過隨機森林（Random Forest）的特徵重要性計算篩選出關鍵特徵，有效提升模型性能並減少冗餘計算。
實驗數據來自美國柏克萊實驗室模擬的空調箱運行數據，涵蓋雙管道與單管道系統的多維度運行參數。實驗結果顯示，本研究所提出的演算法在目標域上的表現明顯優於傳統領域適應方法（如DANN）。在特定故障類型（如冷卻盤管閥門卡住）中，本研究所提出的演算法在Recall指標上達到0.9260，表現優於DANN和其他比較方法。此外，通過結合時間與頻率特徵，本研究顯著提升了目標域的預測準確性，證明了頻率特徵在領域適應中的重要性。
本方法具有多重優勢，包括顯著降低模型對數據標註的依賴、提升無標籤數據的預測能力、以及減少能源浪費的潛力。未來，本系統可應用於國內外高能耗的辦公大樓和工廠，實現暖通空調系統的能效提升與智能化運維，助力「2050淨零排放」目標的實現。

This study addresses the fault diagnosis problem of Air Handling Unit (AHU) systems by proposing an unsupervised learning-based domain adaptation method. It aims to resolve the issue of model performance degradation caused by the data distribution differences between various types of AHUs. The research background highlights the need for improved energy efficiency and the fact that HVAC systems account for 58.62% of electricity consumption in office spaces. To tackle the data distribution inconsistency between Dual Duct Air Handling Units (DDAHU) and Single Duct Air Handling Units (SDAHU), this study utilizes the dual duct system as the source domain and the single duct system as the target domain, enabling label-free cross-domain fault diagnosis.
The proposed method is centered on domain adaptation (DA), employing time-frequency feature encoders to extract data features and leveraging statistical alignment techniques such as Sinkhorn Divergence and Maximum Mean Discrepancy (MMD) to reduce distribution shifts between the domains. During the feature correction stage, an auxiliary decoder is used to reconstruct the original data, thereby optimizing feature representation and improving model accuracy for unlabeled data. Additionally, this study integrates feature importance analysis using Random Forest to identify key features, enhancing model performance while reducing redundant computations.
Experimental data are derived from simulated AHU operation datasets by Lawrence Berkeley National Laboratory, encompassing multi-dimensional operational parameters of dual-duct and single-duct systems. The experimental results demonstrate that our proposal algorithm significantly outperforms traditional domain adaptation methods such as DANN in the target domain. For specific fault types (e.g., stuck cooling coil valve), our proposal algorithm achieved a recall score of 0.9260, surpassing DANN and other comparative methods. Moreover, by incorporating both time and frequency features, this study significantly improves predictive accuracy in the target domain, underscoring the importance of frequency features in domain adaptation.
This approach offers multiple advantages, including a substantial reduction in reliance on labeled data, enhanced prediction capability for unlabeled data, and potential energy savings. In the future, this system could be applied to high-energy-consumption office buildings and factories domestically and internationally, achieving improved HVAC energy efficiency and intelligent operations while contributing to the energy saving and carbon reduction target.

[1] 經濟部能源局2023年非生產性質行業能源查核年報
[2] A. Beghi, R. Brignoli, L. Cecchinato, G. Menegazzo, and M. Rampazzo, "A data-driven approach for fault diagnosis in HVAC chiller systems," in 2015 IEEE Conference on Control Applications (CCA), IEEE, 2015, pp. 966–971. [Online]. Available: https://doi.org/10.1109/CCA.2015.7320737.
[3] D. Chakraborty and H. Elzarka, "Early detection of faults in HVAC systems using an XGBoost model with a dynamic threshold," Energy Build., vol. 185, pp. 326–344, 2019. [Online]. Available: https://doi.org/10.1016/j.enbuild.2018.12.032.
[4] A. Beghi, R. Brignoli, L. Cecchinato, G. Menegazzo, M. Rampazzo, and F. Simmini, "Data-driven fault detection and diagnosis for HVAC water chillers," Contr. Eng. Pract., vol. 53, pp. 79–91, 2016. [Online]. Available: https://doi.org/10.1016/j.conengprac.2016.04.018.
[5] K.W. Roth, D. Westphalen, M.Y. Feng, P. Llana, and L. Quartararo, "Energy Impact of Commercial Building Controls and Performance Diagnostics: Market Characterization, Energy Impact of Building Faults and Energy Savings Potential," Prepared by TAIX LLC for the US Department of Energy, 2005.
[6] N.E. Fernandez, S. Katipamula, W. Wang, Y. Xie, M. Zhao, and C.D. Corbin, "Impacts of Commercial Building Controls on Energy Savings and Peak Load Reduction," PNNL, Richland, WA, 2017. [Online]. Available: https://doi.org/10.2172/1400347.
[7] S. Katipamula and M.R. Brambley, "Methods for fault detection, diagnostics, and prognostics for building systems—a review, part I," HVAC R Res., vol. 11, no. 1, pp. 3–25, 2005. [Online]. Available: https://doi.org/10.1080/10789669.2005.10391123.
[8] Xinyue Huo, Lingxi Xie, Hengtong Hu, Wengang Zhou, Houqiang Li, and Qi Tian, "Domain-agnostic prior for transfer semantic segmentation," arXiv preprint arXiv:2204.02684, 2022.
[9] Yayun Xu and Hua Yan, "Cycle-reconstructive subspace learning with class discriminability for unsupervised domain adaptation," Pattern Recognition, page 108700, 2022.
[10] Anran Zhang, Yandan Yang, Jun Xu, Xianbin Cao, Xiantong Zhen, and Ling Shao, "Latent domain generation for unsupervised domain adaptation object counting," IEEE Transactions on Multimedia, 2022.
[11] Wouter M. Kouw, "An introduction to domain adaptation and transfer learning," arXiv preprint arXiv:1812.11806, 2018.
[12] Artem Rozantsev, Mathieu Salzmann, and Pascal Fua, "Beyond sharing weights for deep domain adaptation," IEEE transactions on pattern analysis and machine intelligence, vol. 41, no. 4, pp. 801–814, 2018.
[13] Baochen Sun, Jiashi Feng, and Kate Saenko, "Return of frustratingly easy domain adaptation," in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 30, 2016.
[14] Guoliang Kang, Lu Jiang, Yi Yang, and Alexander G Hauptmann, "Contrastive adaptation network for unsupervised domain adaptation," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4893–4902, 2019.
[15] Xiaofeng Liu, Yuzhuo Han, Song Bai, Yi Ge, Tianxing Wang, Xu Han, Site Li, Jane You, and Jun Lu, "Importance-aware semantic segmentation in self-driving with discrete wasserstein training," in AAAI, pp. 11629–11636, 2020b.
[16] Junchi Yan, Xu-Cheng Yin, Weiyao Lin, Cheng Deng, Hongyuan Zha, and Xiaokang Yang, "A short survey of recent advances in graph matching," in Proceedings of the 2016 ACM on International Conference on Multimedia Retrieval, pp. 167–174, 2016.
[17] K. M. Borgwardt, A. Gretton, M. J. Rasch, H.-P. Kriegel, B. Schölkopf, and A. J. Smola, "Integrating structured biological data by kernel maximum mean discrepancy," Bioinformatics, vol. 22, no. 14, pp. e49–e57, 2006.
[18] A. Gretton, K. M. Borgwardt, M. J. Rasch, B. Schölkopf, and A. Smola, "A kernel two-sample test," Journal of Machine Learning Research, 2012.
[19] M. Long and J. Wang, "Learning transferable features with deep adaptation networks," CoRR, vol. abs/1502.02791, 1:2, 2015.
[20] H. He, O. Queen, T. Koker, C. Cuevas, T. Tsiligkaridis, and M. Zitnik, "Domain adaptation for time series under feature and label shifts," 2023, arXiv:2302.03133.
[21] Sun, B., Saenko, K., "Deep CORAL: Correlation Alignment for Deep Domain Adaptation," in: Hua, G., Jégou, H. (eds) Computer Vision – ECCV 2016 Workshops. ECCV 2016. Lecture Notes in Computer Science(), vol 9915. Springer, Cham. [Online]. Available: https://doi.org/10.1007/978-3-319-49409-8_35.
[22] Guoliang Kang, Lu Jiang, Yi Yang, Alexander G. Hauptmann, "Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)", 2019, pp. 4893-4902.
[23] N. Courty, R. Flamary, A. Habrard, and A. Rakotomamonjy, "Joint distribution optimal transportation for domain adaptation," in Advances in Neural Information Processing Systems, vol. 30, I. Guyon, U. Von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, Eds. Curran Associates, Inc., 2017.
[24] B. B. Damodaran, B. Kellenberger, R. Flamary, D. Tuia, and N. Courty, "DeepJDOT: Deep joint distribution optimal transport for unsupervised domain adaptation," in Proceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 447–463.
[25] S. Liang, R. Du, and G. Xie, "Model-based fault detection and diagnosis of HVAC systems using support vector machine method," International Journal of Refrigeration, vol. 30, no. 6, pp. 1104–1114, 2007.
[26] H. Wun, J. K.-W. Wong, and S.-H. Ling, "Hybrid random forest and support vector machine modeling for HVAC fault detection and diagnosis," Sensors, vol. 21, no. 8, 2021.
[27] S. Taheri, A. Ahmadi, B. Mohammadi-Ivatloo, and S. Asadi, "Fault detection and diagnosis for HVAC systems via deep learning algorithms," Energy and Buildings, vol. 250, p. 111275, Nov. 2021.
[28] K. Yan, A. Chong, and Y. Mo, "Generative adversarial network for fault detection and diagnosis of chillers," Building and Environment, vol. 172, p. 106698, Apr. 2020.
[29] W.-S. Yun, W.-H. Hong, H. Seo, "A data-driven fault detection and diagnosis scheme for air handling units in building HVAC systems considering undefined states," Journal of Building Engineering, vol. 35, Article no. 102111, 2021, DOI: 10.1016/j.jobe.2020.102111.
[30] Z.-C. Wang, D. Li, Z.-W. Cao, F. Gao, and M.-J. Li, "A modified transformer and adapter-based transfer learning for fault detection and diagnosis in HVAC systems," Energy Storage and Saving, vol. 3, pp. 96–105, Feb. 2024. DOI: 10.1016/j.enss.2024.02.004.
[31] M. Kim, J.-A. Jun, C. Kim, C. S. Pyo, and Y. Song, "Collective AHU Anomaly Detection for Building Energy Optimization," in Proceedings of the International Conference on Information and Communication Technology Convergence (ICTC), 2021, pp. 1-5. doi: 10.1109/ICTC52510.2021.9620913.
[32] Y. Masdoua, M. Boukhnifer, and K. H. Adjallah, "Fault Detection and Diagnosis in AHU System with Data Driven Approaches," in Proceedings of the 8th International Conference on Control, Decision and Information Technologies (CoDIT'22), Istanbul, Turkey, May 17-20, 2022, pp. 1375-1380. doi: 10.1109/CODIT55151.2022.9803907.
[33] Y. Li, Y. Song, L. Jia, S. Gao, Q. Li, and M. Qiu, "Intelligent Fault Diagnosis by Fusing Domain Adversarial Training and Maximum Mean Discrepancy via Ensemble Learning," IEEE Transactions on Industrial Informatics, vol. 17, no. 4, pp. 2833-2841, April 2021. DOI: 10.1109/TII.2020.3008010.
[34] Y. Ganin et al., "Domain-Adversarial Training of Neural Networks," Journal of Machine Learning Research, vol. 17, no. 1, pp. 1-35, 2016.
[35] H. He, O. Queen, T. Koker, C. Cuevas, T. Tsiligkaridis, and M. Zitnik, "Domain Adaptation for Time Series Under Feature and Label Shifts," in Proceedings of the 40th International Conference on Machine Learning, Honolulu, HI, USA, PMLR 202, 2023.
[36] X. Wang, R. Sun, "Domain Adaptation for Time Series Classification," IEEE Access, vol. 12, no. 1, pp. 38456-38472, March 2024, doi: 10.1109/ACCESS.2024.3375334.
[37] H. Yan, Y. Ding, P. Li, Q. Wang, Y. Xu, W. Zuo, "Mind the Class Weight Bias: Weighted Maximum Mean Discrepancy for Unsupervised Domain Adaptation," arXiv preprint arXiv:1705.00609, May 2017. Available: http://arxiv.org/abs/1705.00609v1.
[38] Y. Ozawa, D. Zhao, D. Watari, I. Taniguchi, T. Suzuki, Y. Shimoda, T. Onoye, "Data-driven HVAC Control Using Symbolic Regression: Design and Implementation," in Proceedings of the 2023 IEEE Conference on Decision and Control (CDC), Osaka, Japan, Apr. 2023.
[39] B. Feng, Q. Zhou, J. Xing, Q. Yang, Y. Chen, Z. Deng, "Attention-empowered Transfer Learning Method for HVAC Sensor Fault Diagnosis in Dynamic Building Environments," Building and Environment, vol. 250, Article no. 111148, 2024. Available: https://doi.org/10.1016/j.buildenv.2023.111148.
[40] P. Barrett, J. Hunter, J. T. Miller, J. C. Hsu and P. Greenfield, "Matplotlib - a portable python plotting package," Astronomical Data Analysis Software and Systems XIV, vol. 347, no. 16, pp. 91-95, Dec. 2005.
[41] C. R. Harris, K. J. Millman, S. J. van der Walt, R. Gommers, P. Virtanen, D. Cournapeau, E. Wieser, J. Taylor, S. Berg, N. J. Smith, R. Kern, M. Picus, S. Hoyer, M. H. van Kerkwijk, M. Brett, A. Haldane, J. Fernández del Río, M. Wiebe, P. Peterson, P. Gérard-Marchant, K. Sheppard, T. Reddy, W. Weckesser, H. Abbasi, C. Gohlke and T. E. Oliphant, "Array programming with numpy," Nature, vol. 585, no. 7825, pp. 357 362, Sep. 2020.
[42] F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot and É. Duchesnay, "Scikit-learn: machine learning in python," Journal of Machine Learning Research, vol. 12, no. 85, pp. 2825-2830, Oct. 2011.
[43] Lawrence Berkeley National Laboratory, LBNL FDD Data Sets. DOI: https://dx.doi.org/10.25984/188132
[44] 新北市政府「空調系統節能管理與維護手冊」
[45] Yun, W.-S., Hong, W.-H., and Seo, H., "A data-driven fault detection and diagnosis scheme for air handling units in building HVAC systems considering undefined states," Journal of Building Engineering, vol. 35, p. 102111, 2021, doi: 10.1016/j.jobe.2020.102111.