Bridging The Synthetic-To-Real Gap: A Model-Data Coevolution Approach With Stochastic Feature Decoupling  For Ac Unit Fault Diagnosis

Muhamad Raynard Alif; Mukhammad Andri Setiawan

doi:10.61132/mars.v3i6.1262

Authors

Muhamad Raynard Alif Universitas Islam Indonesia
Mukhammad Andri Setiawan Universitas Islam Indonesia

DOI:

https://doi.org/10.61132/mars.v3i6.1262

Keywords:

Model-Data Coevolution, Synthetic Data, Domain Gap, Sim-to-Real, Fault Detection and Diagnosis (FDD)

Abstract

The scarcity of real-world data in Air-Conditioning (AC) fault diagnosis necessitates the use of synthetic data; however, rule-based synthetic datasets often suffer from a significant sim-to-real domain gap. To address this, we propose a Model-Data Coevolution (MDC) framework that employs a Simulated Annealing (SA) controller to optimize augmentation parameters. We introduce a novel technique, Stochastic Feature Decoupling (SFD), which applies independent noise to raw and derived features, contrasting it with traditional Logically-Consistent Augmentation (LCA). Empirical results show that SFD significantly outperforms LCA, achieving a weighted F1-score of 0.93 and increasing NORMAL class recall to 82%. We demonstrate that by breaking deterministic feature links, SFD acts as a robust regularizer, utilizing "physically impossible" data to enhance generalization in complex real-world environments.

References

Matetić, I., Štajduhar, I., Wolf, I. and Ljubić, S. (2022) 'A Review of Data-Driven Approaches and Techniques for Fault Detection and Diagnosis in HVAC Systems', Sensors, 23(1), Art. no. 1. doi: 10.3390/s23010001.

Aghili, S. A., Rezaei, A. H. M., Tafazzoli, M., Khanzadi, M. and Rahbar, M. (2025) 'Artificial Intelligence Approaches to Energy Management in HVAC Systems: A Systematic Review', Buildings, 15(7), Art. no. 1008. doi: 10.3390/buildings15071008.

Allen-Magande, H., Khazaii, J. and Esmaeili, A. (2023) 'A Literature Review of Automated Fault Detection and Diagnostics for HVAC Systems', in ASME 2023 International Mechanical Engineering Congress and Exposition (IMECE), New Orleans, LA, USA, Paper no. IMECE2023-111611. doi: 10.1115/IMECE2023-111611.

Zhang, J., Zhang, C., Lu, J. and Zhao, Y. (2025) 'Domain-specific large language models for fault diagnosis of heating, ventilation, and air conditioning systems by labeled-data-supervised fine-tuning', Applied Energy, 377, Art. no. 124378. doi: 10.1016/j.apenergy.2024.124378.

Crowe, E. et al. (2023) 'Empirical analysis of the prevalence of HVAC faults in commercial buildings', Science and Technology for the Built Environment, 29(10), pp. 1027–1038. doi: 10.1080/23744731.2023.2263324.

Yan, K., Yang, B. and Zhuang, C. (2025) 'Generative Models for HVAC Fault Detection and Diagnosis in Indoor and Built Environment', Journal of Building Design and Environment, 3, Art. no. 202556. doi: 10.70401/jbde.2025.0012.

Abdollah, M. A. F. (2024) Data driven Fault Detection and Diagnostics for HVAC systems in buildings. Ph.D. dissertation. Milan, Italy: Departement of Energy, Politecnico di Milano.

Sharma, V. and Mistry, V. (2023) Automated Fault Detection and Diagnostics in HVAC Systems. Zenodo. doi: 10.5281/zenodo.11079964.

Singh, V., Mathur, J. and Bathia, A. (2022) 'A comprehensive review of fault detection, diagnostics, prognostics, and fault modelling in HVAC systems', International Journal of Refrigeration, 144, pp. 142–162. doi: 10.1016/j.ijrefrig.2022.08.017.

Chen, Z. et al. (2023) 'A review of data-driven fault detection and diagnostics for building HVAC systems', Applied Energy, 339, Art. no. 121030. doi: 10.1016/j.apenergy.2023.121030.

Li, G. et al. (2021) 'Review on Fault Detection and Diagnosis Feature Engineering in Building Heating, Ventilation, Air Conditioning and Refrigeration Systems', IEEE Access, 9, pp. 2153-2187. doi: 10.1109/ACCESS.2020.3040980.

Kašiković, M. and Anđelković, A. (2025) 'Fault detection and diagnosis methods in HVAC systems', in 55th International HVAC&R Congress and Exhibition. doi: 10.24094/kghk.024.1.075.

Wu, R., Ren, Y., Tan, M. and Nie, L. (2024) 'Fault diagnosis of HVAC system with imbalanced data using multi-scale convolution composite neural network', Building Simulation, 17(3), pp. 1–16. doi: 10.1007/s12273-023-1086-1.

Gao, X. et al. (2025) 'Class-imbalanced domain generalization fault diagnosis for chiller based on simulation data', Science and Technology for the Built Environment, pp. 1–14. doi: 10.1080/23744731.2025.2556955.

Chen, H. et al. (2024) 'Review of imbalanced fault diagnosis technology based on generative adversarial networks', Journal of Computational Design and Engineering, 11(5), pp. 99–124. doi: https://doi.org/10.1093/jcde/qwae075.

Granderson, J. et al. (2023) 'A labeled dataset for building HVAC systems operating in faulted and fault-free states', Scientific Data, 10, Art. no. 342. doi: https://doi.org/10.1038/s41597-023-02197-w.

Sol, J. et al. (2024) 'Sim-to-Real Domain Adaptation for Deformation Classification', in 2024 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Kuching, Malaysia, pp. 2225-2231. doi: 10.1109/SMC54092.2024.10831103.

Mikołajewska, E. et al. (2025) 'Generative AI in AI-Based Digital Twins for Fault Diagnosis for Predictive Maintenance in Industry 4.0/5.0', Appl. Sci., 15(6), Art. no. 3166. doi: 10.3390/app15063166.

Deng, Z. et al. (2025) 'Research on equipment fault diagnosis model based on gan and inverse PINN: Solutions for data imbalance and rare faults', PLOS ONE, 20(5), Art. no. e0324180. doi: 10.1371/journal.pone.0324180.

Forsberg, F. (2021) Domain Adaptation to meet the Reality-Gap from Simulation to Reality. Master of Science Thesis. Linköping, Sweden: Department of Electrical Engineering, Linköping University.

Ugurlu, H. I., Pham, X. H. and Kayacan, E. (2022) 'Sim-to-Real Deep Reinforcement Learning for Safe End-to-End Planning of Aerial Robots', Robotics, 11(5), Art. no. 109. doi: 10.3390/robotics11050109.

Rizzardo, C., Chen, F. and Caldwell, D. (2023) 'Sim-to-real via latent prediction: Transferring visual non-prehensile manipulation policies', Front. Robot. AI, 9, Art. no. 1067502. doi: 10.3389/frobt.2022.1067502.

Chen, R. J. et al. (2021) 'Synthetic data in machine learning for medicine and healthcare', Nat. Biomed. Eng., 5(6), pp. 493–497. doi: 10.1038/s41551-021-00751-8.

Alberts, M. et al. (2024) 'Transitioning from Simulation to Reality: Applying Chatter Detection Models to Real-World Machining Data', Machines, 12(12), Art. no. 923. doi: 10.3390/machines12120923.

Gordienko, Y. et al. (2025) 'Effect of natural and synthetic noise data augmentation on physical action classification by brain– computer interface and deep learning', Front. Neuroinform., 19, Art. no. 1521805. doi: 10.3389/fninf.2025.1521805.

Joshi, S. et al. (2025) 'Towards Mitigating Spurious Correlations in the Wild: A Benchmark and a more Realistic Dataset', arXiv preprint arXiv:2306.11957. doi: 10.48550/arXiv.2306.11957.