Keywords
noisy time series
diffusion model
anomaly detection
noise disentanglement
equipment fault diagnosis
time-frequency analysis
Abstract
Industrial acoustic monitoring provides non-contact condition information for motors, pumps, compressors, cutting machines, fans, and rotating equipment. However, acoustic time series are strongly contaminated by background noise, multi-machine interference, airflow disturbance, workshop vibration, and transient operational sounds. These factors make it difficult to detect early abnormal acoustic patterns associated with mechanical degradation. This study proposes a noise-disentangled diffusion modeling approach for abnormal acoustic pattern detection in industrial equipment. The method converts acoustic streams into time-frequency sequences and uses a conditional diffusion model to reconstruct clean latent acoustic patterns from noisy spectral observations. A disentanglement constraint separates background workshop noise, normal machine-state variation, and fault-related acoustic components. Anomaly scores are computed from denoised spectral residuals and fault-component activation intensity. Experiments are conducted on an industrial acoustic dataset collected from 1,420 machines across 16 workshops, including motors, centrifugal pumps, air compressors, and spindle systems. The dataset contains 9,600 hours of acoustic recordings, 5.4 million time-frequency windows, and 1,180 maintenance-confirmed abnormal episodes, including bearing wear, pump cavitation, belt misalignment, compressor leakage, and spindle chatter. Compared with a spectral autoencoder baseline, the proposed method reduces median abnormal-sound detection delay from 31.5 minutes to 8.2 minutes. False alerts decrease to 1.3 cases per machine-month in high-noise workshops. The average denoised spectral distortion is reduced by 0.112 log-magnitude units, and fault-related component separation lowers background-noise activation by 6.9 dB. The model processes 26,000 acoustic windows per second during offline assessment and maintains 52 ms median inference latency for online monitoring. The results demonstrate that diffusion-driven denoising and component disentanglement can improve robust anomaly detection for noisy industrial acoustic time series.
References
Xiong, W., Zeng, Y., & Guo, Y. (2026). A Study on the Impact of MEP Space Organization in Large-Scale Commercial Complexes on Investment Efficiency and Its Mechanisms.
Rychlik, A. (2025). Technical Condition Assessment of Light-Alloy Wheel Rims Based on Acoustic Parameter Analysis Using a Neural Network. Sensors, 25(14), 4473.
Hong, Z., Xu, T., & Chen, H. (2026). A Study on Test Coverage Mapping and Release Risk Prediction in Continuous Delivery of Cloud Services.
Johnson, J. C., & Rong, Y. (2026). Automated Classification of Humpback Whale Calls Using Deep Learning: A Comparative Study of Neural Architectures and Acoustic Feature Representations. Sensors, 26(2), 715.
Chen, X., Xiao, H., Zeng, Z., Zhang, S., & Xiao, R. (2025). Fine-Grained Multivariate Time Series Anomaly Detection via Causal Inference. Knowledge-Based Systems, 114765.
Pichler, C., Neumayer, M., Schweighofer, B., Feilmayr, C., Schuster, S., & Wegleiter, H. (2025). Decomposing and modeling acoustic signals to identify machinery defects in industrial soundscapes. Sensors, 25(16), 4923.
Li, J., Ma, R., & Gu, Y. (2026). From Audit Requirements to Computable Software Architecture: A Rule-Engine and Evidence-Indexing Method for Trusted Digital Infrastructure.
Fu, Y., Gui, H., Li, W., & Wang, Z. (2020, August). Virtual Material Modeling and Vibration Reduction Design of Electron Beam Imaging System. In 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications (AEECA) (pp. 1063-1070). IEEE.
Anwar, T., Mu, C., Yousaf, M. Z., Khan, W., Khalid, S., Hourani, A. O., & Zaitsev, I. (2025). Robust fault detection and classification in power transmission lines via ensemble machine learning models. Scientific Reports, 15(1), 2549.
Chen, H., Ma, X., Mao, Y., & Ning, P. (2025). Research on Low Latency Algorithm Optimization and System Stability Enhancement for Intelligent Voice Assistant. Available at SSRN 5321721.
Liang, R., Feifan, F. N. U., Liang, Y., & Ye, Z. (2025). Emotion-Aware Interface Adaptation in Mobile Applications Based on Color Psychology and Multimodal User State Recognition. Frontiers in Artificial Intelligence Research, 2(1), 51-57.
Moorthy, S., & Moon, Y. K. (2025). Hybrid multi-attention network for audio–visual emotion recognition through multimodal feature fusion. Mathematics, 13(7), 1100.
Yin, J., Huang, Y., & Rao, H. (2026). A Study on the Dynamic Evolution of Learning Behavior and Outcome Prediction in Digital Educational Environments. Available at SSRN 6607139.
Yang, J. (2026). Computational Analysis of How Digital Non-Clinical Communication Tools Influence Social Participation Among Older Adults in Community-Based Elderly Care. Available at SSRN 6682838.
Farhadi, S., Kohlbacher, M., & Lienhart, W. (2026). Iterative self-training segmentation for scalable traffic monitoring using distributed acoustic sensing. Computer-Aided Civil and Infrastructure Engineering, 100136.
Zhang, Z. (2026). A Study on Multimodal Product Understanding and Assembly Guidance Optimization in e-commerce for Complex Consumer Goods. Available at SSRN 6734559.
Zhang, Z., Gao, Y., & Tong, Y. (2026). Semantic-Temporal Graph Learning for Demand Forecasting and Resource Allocation in Public Information Systems. Available at SSRN 6792279.
Maliuk, A., Nguyen, T. K., Ahmad, Z., & Kim, J. M. (2026). KFD-AEEScan: a Katz fractal dimension acoustic emission event scanner for single-sensor multi-fault diagnosis in milling machines. Structural Health Monitoring, 14759217261433437.
Qi, C., & Qiao, X. (2026). Building Reliable AI Systems: A Framework That Bridges Architecture and Operations. Available at SSRN 6795298.
Yin, L., Tong, Y., Xie, R., Zhang, Z., Islam, Z. H., Zhang, K., ... & Wang, B. (2024). Targeted NAD+ repletion via biomimetic nanoparticle enables simultaneous management of intimal hyperplasia and accelerated re-endothelialization: a proof-of-concept study toward next-generation of endothelium-protective, anti-restenotic therapy. Journal of Controlled Release, 376, 806-815.
Szymoniak, S., & Kuczyński, Ł. (2025). Overview of modern technologies for acquiring and analysing acoustic information based on AI and IoT. Applied Sciences, 15(12), 6690.
Zhang, Y., Gu, W., & Wang, J. (2025, December). Construction of Wind Farm Asset Health Index Based on Multi-Dimensional Indicators and Analytic Hierarchy Process and Its Correlation with Operational Performance. In Proceedings of the 2025 International Conference on Digital Transformation and Management (pp. 317-323).
Su, D., & Dong, Y. (2026). Classroom-Based Assessment with Bayesian Learning Analytics for Instructional Decision-Making in ASD Inclusive Education.
Idrees, M., Huang, Y., & Li, A. (2026). Latent-optimized collaborative feature disentanglement for enhanced shadow removal. Pattern Analysis and Applications, 29(3), 132.
Xu, T., Zhang, J., & Zhu, W. (2026). Reproducible Modeling Pipelines and Cross-Window Stability in Subprime Auto Loan Credit Risk Assessment. Available at SSRN 6893861.
Xie, Z., Ren, X., Zheng, T., Bai, J., Fan, W., Xu, B., ... & Song, Y. (2026). A Survey on AI Agent Harness. ResearchGate Preprint. DOI, 10.
Rohan, A. I., Ridita, T. A., Anonto, H. Z., Hossain, M. I., Shufian, A., Mahin, M. S. R., & Islam, M. A. (2025). Intelligent fault diagnosis in rolling element bearings: Combining envelope spectrum and spectral kurtosis for enhanced detection. Results in Engineering, 106899.
Jeong, S., Kim, H., Kim, Y. H., Park, C. S., Jung, H., & Kim, H. K. (2025). Spatiotemporal anomaly detection in distributed acoustic sensing using a GraphDiffusion model. Sensors, 25(16), 5157.
Zhang, Y., Gu, W., & Wang, J. (2025, December). Research on First Article Inspection (FAI)-Driven Quality Assurance Methods for Wind Turbine Installation and Operation & Maintenance and Their Effect on Reliability Improvement. In Proceedings of the 2025 6th International Conference on Computer Science and Management Technology (pp. 1700-1705).
Yin, L., Tong, Y., Islam, Z. H., Zhang, K., Xie, R., Burger, J., ... & Wang, B. (2024). Targeted NAD+ Delivery for intimal Hyperplasia and Re-endothelialization: a novel anti-restenotic therapy approach. bioRxiv, 2024-02.
Antony, A. S. M., Sundaram, K. M., Raman, C. J., & Murthy, G. R. (2026). Intelligent fault detection in battery systems: a machine learning approach with transformer-enhanced multi-modal Sensing. Electric Power Systems Research, 251, 112279.
Du, Y., Liu, Q., Dong, Y., & Chen, Z. (2026). Cross-Domain Transfer and Few-Shot Recognition of Defect Images in Complex Industrial Settings.