This is a preview of subscription content, log in via an institution to check access.
References
Liang G L, Ma W, Wang Y L. Time-space transform: a novel signal processing approach for an acoustic vector-sensor. Sci China Inf Sci, 2013, 56: 042313
Lympertos E M, Dermatas E S. Acoustic emission source location in dispersive media. Signal Processing, 2007, 87: 3218–3225
Sousa K M, Dreyer U J, Martelli C, et al. Dynamic eccentricity induced in induction motor detected by optical fiber Bragg grating strain sensors. IEEE Sens J, 2016, 16: 4786–4792
Zhang W T, Jiang J W, Shao Y X, et al. Snapshot boosting: a fast ensemble framework for deep neural networks. Sci China Inf Sci, 2020, 63: 112102
Liu Z Z, Zhang X Z, Jiang J F, et al. Stabilization of high sensitivity optical fiber AE sensing for long-term detection. Optical Fiber Tech, 2021, 61: 102391
Wu Q, Yu F M, Okabe Y, et al. Application of a novel optical fiber sensor to detection of acoustic emissions by various damages in CFRP laminates. Smart Mater Struct, 2015, 24: 015011
Acknowledgements
This work was supported by National Natural Science Foundation of China (Grant Nos. U1833104, 61735011).
Author information
Authors and Affiliations
Corresponding authors
Additional information
Conclusion
In this study, the problem of inadequate or uneven data collection is studied. A small sample dataset tends to cause overfitting of deep learning models, which limits the application of deep neural networks in engineering. To overcome this problem, data augmentation methods of random scale-cropping as well as random erasing are proposed. The results show that with the combination of the above methods, the model exhibits excellent classification performance with an accuracy of 90.46%. Further, the data augmentation methods proposed in the study have the potential to become general solutions in many fields besides fiber sensing, which guarantees that deep learning models can be effectively applied in engineering practices.
Supporting information
Appendixes A and B. The supporting information is available online at info.scichina.com and link.springer.com. The supporting materials are published as submitted, without typesetting or editing. The responsibility for scientific accuracy and content remains entirely with the authors.
Rights and permissions
About this article
Cite this article
Zhang, X., Sun, H., Jiang, J. et al. Optical time-series signals classification based on data augmentation for small sample. Sci. China Inf. Sci. 65, 229303 (2022). https://doi.org/10.1007/s11432-022-3615-1
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11432-022-3615-1