{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,16]],"date-time":"2026-07-16T04:04:41Z","timestamp":1784174681539,"version":"3.55.0"},"reference-count":50,"publisher":"MDPI AG","issue":"14","license":[{"start":{"date-parts":[[2026,7,15]],"date-time":"2026-07-15T00:00:00Z","timestamp":1784073600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Magdeburg University"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>Vibration-based mechanical fault diagnosis has become a critical research area, mostly driven by the need to improve equipment reliability and reduce unplanned downtime in industrial settings. Time\u2013Frequency Networks (TFNs) have shown strong potential here, combining interpretable time\u2013frequency transformations with deep learning classifiers in a single framework. This work reproduces the original TFN model from the recent literature and evaluates its noise robustness under additive Gaussian noise (10 dB, 0 dB, \u22125 dB SNR) and impulsive noise at the same levels, across five architectures: Backbone CNN, Random CNN, TFN-Chirplet, TFN-Morlet, and a squeeze-and-excitation attention CNN baseline. The evaluation protocol corrects two methodological issues identified during peer review of an earlier version of this work\u2014window-level data leakage between train and test splits, and selection of the best-performing training epoch rather than a fixed final-epoch result\u2014both of which are shown to materially affect reported outcomes. Under the corrected protocol, TFN-Morlet remains the most noise-robust architecture, with only a 19.09% accuracy drop from clean to \u22125 dB AWGN, approximately 15.5 percentage points better than Backbone CNN under the same conditions; an architectural anomaly reported in the earlier version of this study, in which mild noise appeared to improve an unconstrained CNN\u2019s accuracy, was not reproduced under the corrected protocol and is shown to be an artifact of the original methodological issues. Per-class analysis and multi-model confusion matrices further reveal that misclassifications under severe noise are dominated by confusion between the same defect severity at different fault locations, rather than between different severities at the same location as previously reported. These results indicate that time\u2013frequency-aware convolutional kernels improve both classification accuracy and noise resistance under rigorous, leakage-free evaluation, and that this robustness is not replicated by a generic attention mechanism alone.<\/jats:p>","DOI":"10.3390\/s26144492","type":"journal-article","created":{"date-parts":[[2026,7,16]],"date-time":"2026-07-16T03:34:23Z","timestamp":1784172863000},"page":"4492","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Noise Robustness Evaluation of Time\u2013Frequency Networks (TFNs) for Intelligent Mechanical Fault Diagnosis"],"prefix":"10.3390","volume":"26","author":[{"given":"Syed Khizar","family":"Zubair","sequence":"first","affiliation":[{"name":"College of Electrical and Mechanical Engineering (CEME), National University of Sciences and Technology (NUST), Islamabad 46020, Pakistan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Imran","family":"Shafi","sequence":"additional","affiliation":[{"name":"College of Electrical and Mechanical Engineering (CEME), National University of Sciences and Technology (NUST), Islamabad 46020, Pakistan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2788-2659","authenticated-orcid":false,"given":"Ahmet","family":"Caglar","sequence":"additional","affiliation":[{"name":"Department of Mechanical Engineering, Akdeniz University, 07070 Antalya, Turkey"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Abdul Saboor","family":"Khan","sequence":"additional","affiliation":[{"name":"Department of Electrical Engineering and Information Technology, Otto-von-Guericke University, 39106 Magdeburg, Germany"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jamil","family":"Ahmad","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Abdul Wali Khan University, Mardan 23200, Pakistan"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,7,15]]},"reference":[{"key":"ref_1","first-page":"6501309","article-title":"Rolling bearing fault detection based on improved piecewise unsaturated bistable stochastic resonance method","volume":"70","author":"Li","year":"2021","journal-title":"IEEE Trans. Instrum. Meas."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"157","DOI":"10.1016\/j.jmsy.2018.01.006","article-title":"Data-driven smart manufacturing","volume":"48","author":"Tao","year":"2018","journal-title":"J. Manuf. Syst."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"1700","DOI":"10.1109\/TITS.2020.3029946","article-title":"Data-driven fault diagnosis for traction systems in high-speed trains: A survey, challenges, and perspectives","volume":"23","author":"Chen","year":"2022","journal-title":"IEEE Trans. Intell. Transp. Syst."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"106587","DOI":"10.1016\/j.ymssp.2019.106587","article-title":"Applications of machine learning to machine fault diagnosis: A review and roadmap","volume":"138","author":"Lei","year":"2020","journal-title":"Mech. Syst. Signal Process."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"990","DOI":"10.1109\/TIM.2011.2179819","article-title":"Rotational machine health monitoring and fault detection using EMD-based acoustic emission feature quantification","volume":"61","author":"Li","year":"2012","journal-title":"IEEE Trans. Instrum. Meas."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"330","DOI":"10.1016\/j.measurement.2014.12.021","article-title":"An improvement EMD method based on the optimized rational Hermite interpolation approach and its application to gear fault diagnosis","volume":"63","author":"Li","year":"2015","journal-title":"Measurement"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"199","DOI":"10.1016\/S0888-3270(03)00075-X","article-title":"Application of the wavelet transform in machine condition monitoring and fault diagnostics: A review with bibliography","volume":"18","author":"Peng","year":"2004","journal-title":"Mech. Syst. Signal Process."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"6506","DOI":"10.1109\/TIE.2017.2650873","article-title":"Application of bandwidth EMD and adaptive multiscale morphology analysis for incipient fault diagnosis of rolling bearings","volume":"64","author":"Li","year":"2017","journal-title":"IEEE Trans. Ind. Electron."},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"2626","DOI":"10.1109\/TIE.2020.2972472","article-title":"Key-performance-indicator-related process monitoring based on improved kernel partial least squares","volume":"68","author":"Si","year":"2021","journal-title":"IEEE Trans. Ind. Electron."},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"2727","DOI":"10.1109\/TIE.2017.2745473","article-title":"Electric locomotive bearing fault diagnosis using a novel convolutional deep belief network","volume":"65","author":"Shao","year":"2018","journal-title":"IEEE Trans. Ind. Electron."},{"key":"ref_11","first-page":"6114","article-title":"Coarse-to-fine: Progressive knowledge transfer-based multitask convolutional neural network for intelligent large-scale fault diagnosis","volume":"33","author":"Wang","year":"2021","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"6339","DOI":"10.1109\/TNNLS.2021.3135877","article-title":"Intelligent fault diagnosis of gearbox under variable working conditions with adaptive intraclass and interclass convolutional neural network","volume":"34","author":"Zhao","year":"2022","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"17","DOI":"10.1016\/j.neucom.2018.01.092","article-title":"Computer vision and deep learning techniques for pedestrian detection and tracking: A survey","volume":"300","author":"Brunetti","year":"2018","journal-title":"Neurocomputing"},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"60","DOI":"10.1016\/j.neunet.2017.02.013","article-title":"Evaluating deep learning architectures for speech emotion recognition","volume":"92","author":"Fayek","year":"2017","journal-title":"Neural Netw."},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"354","DOI":"10.1038\/nature24270","article-title":"Mastering the game of Go without human knowledge","volume":"550","author":"Silver","year":"2017","journal-title":"Nature"},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"27","DOI":"10.1631\/FITEE.1700808","article-title":"Visual interpretability for deep learning: A survey","volume":"19","author":"Zhang","year":"2018","journal-title":"Front. Inf. Technol. Electron. Eng."},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"110952","DOI":"10.1016\/j.ymssp.2023.110952","article-title":"TFN: An interpretable neural network with time-frequency transform embedded for intelligent fault diagnosis","volume":"207","author":"Chen","year":"2024","journal-title":"Mech. Syst. Signal Process."},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"2302","DOI":"10.1109\/TSMC.2020.3048950","article-title":"WaveletKernelNet: An interpretable deep neural network for industrial intelligent diagnosis","volume":"52","author":"Li","year":"2022","journal-title":"IEEE Trans. Syst. Man Cybern. Syst."},{"key":"ref_19","unstructured":"Oppenheim, A.V., Willsky, A.S., Nawab, S.H., and Hern\u00e1ndez, G.M. (1997). Signals & Systems, Pearson Educaci\u00f3n."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"4949","DOI":"10.1109\/TII.2020.2967557","article-title":"Multibranch and multiscale CNN for fault diagnosis of wheelset bearings under strong noise and variable load condition","volume":"16","author":"Peng","year":"2020","journal-title":"IEEE Trans. Ind. Informat."},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"224","DOI":"10.1016\/j.isatra.2020.08.010","article-title":"Deep learning algorithms for rotating machinery intelligent diagnosis: An open source benchmark study","volume":"107","author":"Zhao","year":"2020","journal-title":"ISA Trans."},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"56","DOI":"10.1016\/j.ast.2018.08.042","article-title":"Least squares support vector machine for class imbalance learning and their applications to fault detection of aircraft engine","volume":"84","author":"Xi","year":"2019","journal-title":"Aerosp. Sci. Technol."},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"1271","DOI":"10.1007\/s10845-020-01608-8","article-title":"A novel normalized recurrent neural network for fault diagnosis with noisy labels","volume":"32","author":"Nie","year":"2021","journal-title":"J. Intell. Manuf."},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"228","DOI":"10.1016\/j.patrec.2021.06.030","article-title":"Perturbation-based methods for explaining deep neural networks: A survey","volume":"150","author":"Ivanovs","year":"2021","journal-title":"Pattern Recognit. Lett."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"741","DOI":"10.1109\/TRPMS.2021.3066428","article-title":"On interpretability of artificial neural networks: A survey","volume":"5","author":"Fan","year":"2021","journal-title":"IEEE Trans. Radiat. Plasma Med. Sci."},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"726","DOI":"10.1109\/TETCI.2021.3100641","article-title":"A survey on neural network interpretability","volume":"5","author":"Zhang","year":"2021","journal-title":"IEEE Trans. Emerg. Top. Comput. Intell."},{"key":"ref_27","unstructured":"Dhurandhar, A., Chen, P.-Y., Luss, R., Tu, C.-C., Ting, P., Shanmugam, K., and Das, P. (2018). Explanations based on the missing: Towards contrastive explanations with pertinent negatives. Proceedings of Advances in Neural Information Processing Systems, Curran Associates, Inc."},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"Wang, Y., Su, H., Zhang, B., and Hu, X. (2018, January 18\u201323). Interpret neural networks by identifying critical data routing paths. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA.","DOI":"10.1109\/CVPR.2018.00928"},{"key":"ref_29","doi-asserted-by":"crossref","unstructured":"Bau, D., Zhou, B., Khosla, A., Oliva, A., and Torralba, A. (2017, January 21\u201326). Network dissection: Quantifying interpretability of deep visual representations. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA.","DOI":"10.1109\/CVPR.2017.354"},{"key":"ref_30","unstructured":"Simonyan, K., Vedaldi, A., and Zisserman, A. (2014). Deep inside convolutional networks: Visualising image classification models and saliency maps. arXiv."},{"key":"ref_31","doi-asserted-by":"crossref","unstructured":"Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., and Batra, D. (2017, January 22\u201329). Grad-CAM: Visual explanations from deep networks via gradient-based localization. Proceedings of the IEEE International Conference on Computer Vision, Venice, Italy.","DOI":"10.1109\/ICCV.2017.74"},{"key":"ref_32","unstructured":"Sundararajan, M., Taly, A., and Yan, Q. (2017, January 6\u201311). Axiomatic attribution for deep networks. Proceedings of the 34th International Conference on Machine Learning, Sydney, Australia."},{"key":"ref_33","doi-asserted-by":"crossref","unstructured":"Li, O., Liu, H., Chen, C., and Rudin, C. (2018). Deep learning for case-based reasoning through prototypes: A neural network that explains its predictions. Proceedings of the AAAI Conference on Artificial Intelligence, AAAI Press.","DOI":"10.1609\/aaai.v32i1.11771"},{"key":"ref_34","unstructured":"Ravanelli, M., and Bengio, Y. (2019). Interpretable convolutional filters with SincNet. arXiv."},{"key":"ref_35","first-page":"1831","article-title":"Wavelet kernel based convolutional neural network for localization of partial discharge sources within a power apparatus","volume":"17","author":"Ganguly","year":"2020","journal-title":"IEEE Trans. Ind. Informat."},{"key":"ref_36","doi-asserted-by":"crossref","first-page":"5845","DOI":"10.1109\/TNNLS.2021.3071564","article-title":"Whitening-Net: A generalized network to diagnose the faults among different machines and conditions","volume":"33","author":"Li","year":"2021","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"ref_37","doi-asserted-by":"crossref","first-page":"107327","DOI":"10.1016\/j.ymssp.2020.107327","article-title":"A hybrid classification autoencoder for semi-supervised fault diagnosis in rotating machinery","volume":"149","author":"Wu","year":"2021","journal-title":"Mech. Syst. Signal Process."},{"key":"ref_38","doi-asserted-by":"crossref","first-page":"108673","DOI":"10.1016\/j.ymssp.2021.108673","article-title":"Fully interpretable neural network for locating resonance frequency bands for machine condition monitoring","volume":"168","author":"Wang","year":"2022","journal-title":"Mech. Syst. Signal Process."},{"key":"ref_39","doi-asserted-by":"crossref","first-page":"44","DOI":"10.1186\/s10033-021-00564-5","article-title":"An interpretable denoising layer for neural networks based on reproducing kernel Hilbert space and its application in machine fault diagnosis","volume":"34","author":"Zhao","year":"2021","journal-title":"Chin. J. Mech. Eng."},{"key":"ref_40","doi-asserted-by":"crossref","first-page":"e2106598119","DOI":"10.1073\/pnas.2106598119","article-title":"Fully learnable deep wavelet transform for unsupervised monitoring of high-frequency time series","volume":"119","author":"Michau","year":"2022","journal-title":"Proc. Natl. Acad. Sci. USA"},{"key":"ref_41","doi-asserted-by":"crossref","unstructured":"Hakim, M., Omran, A.A.B., Inayat-Hussain, J.I., Ahmed, A.N., Abdellatef, H., Abdellatif, A., and Gheni, H.M. (2022). Bearing fault diagnosis using lightweight and robust one-dimensional convolution neural network in the frequency domain. Sensors, 22.","DOI":"10.3390\/s22155793"},{"key":"ref_42","doi-asserted-by":"crossref","first-page":"977","DOI":"10.1038\/s41598-025-85275-w","article-title":"Attention activation network for bearing fault diagnosis under various noise environments","volume":"15","author":"Zhang","year":"2025","journal-title":"Sci. Rep."},{"key":"ref_43","doi-asserted-by":"crossref","unstructured":"Liu, H., Zhang, F., Tan, Y., Huang, L., Li, Y., Huang, G., Luo, S., and Zeng, A. (2024). Multi-scale quaternion CNN and BiGRU with cross self-attention feature fusion for fault diagnosis of bearing. arXiv.","DOI":"10.1088\/1361-6501\/ad4c8e"},{"key":"ref_44","doi-asserted-by":"crossref","unstructured":"Hu, W., Zhou, H., and Yang, J. (2026). Interpretable and noise-robust bearing fault diagnosis for CNC machine tools via adaptive shapelet-based deep learning model. Machines, 14.","DOI":"10.3390\/machines14020214"},{"key":"ref_45","doi-asserted-by":"crossref","unstructured":"Hu, J., Shen, L., and Sun, G. (2018, January 18\u201323). Squeeze-and-excitation networks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA.","DOI":"10.1109\/CVPR.2018.00745"},{"key":"ref_46","doi-asserted-by":"crossref","first-page":"3948","DOI":"10.1109\/TIE.2012.2206331","article-title":"Multicomponent signal analysis based on polynomial chirplet transform","volume":"60","author":"Yang","year":"2013","journal-title":"IEEE Trans. Ind. Electron."},{"key":"ref_47","doi-asserted-by":"crossref","first-page":"115571","DOI":"10.1016\/j.jsv.2020.115571","article-title":"Iterative nonlinear chirp mode decomposition: A Hilbert-Huang transform-like method in capturing intra-wave modulations of nonlinear responses","volume":"485","author":"Tu","year":"2020","journal-title":"J. Sound Vib."},{"key":"ref_48","doi-asserted-by":"crossref","first-page":"81","DOI":"10.1016\/j.neuroimage.2019.05.048","article-title":"A better way to define and describe Morlet wavelets for time-frequency analysis","volume":"199","author":"Cohen","year":"2019","journal-title":"NeuroImage"},{"key":"ref_49","doi-asserted-by":"crossref","first-page":"63","DOI":"10.1016\/j.patrec.2016.08.016","article-title":"Using filter banks in convolutional neural networks for texture classification","volume":"84","author":"Andrearczyk","year":"2016","journal-title":"Pattern Recognit. Lett."},{"key":"ref_50","unstructured":"Kingma, D.P., and Ba, J. (2015, January 7\u20139). Adam: A method for stochastic optimization. Proceedings of the International Conference on Learning Representations (ICLR), San Diego, CA, USA."}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/14\/4492\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,16]],"date-time":"2026-07-16T03:37:54Z","timestamp":1784173074000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/14\/4492"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,15]]},"references-count":50,"journal-issue":{"issue":"14","published-online":{"date-parts":[[2026,7]]}},"alternative-id":["s26144492"],"URL":"https:\/\/doi.org\/10.3390\/s26144492","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,15]]}}}