Abstract

Effective information mining of fiber-optic distributed acoustic sensors (DAS) is so important that it attracts more and more public attention, and various manual and deep feature extraction methods have been developed. However, either way it has limits; for example, the manual features contain insufficient information, and the deep features could be unreliable because of the overfitting problem. Thus, in this paper, to avoid the disadvantages of each and make full use of the effective information carried by DAS signals, an intelligent target recognition method by utilizing both manual and deep features is proposed. The manual features are first extracted in the time domain, frequency domain, semantic domain, and from dynamic models, which are fused with the deep features extracted by a four-layer 1D convolutional neural network (CNN) through feature engineering. The features are ranked and then selected by a combined weighting method of analysis of variance and maximum information coefficient. Then finally, an optimal classifier is selected by comparing support vector machine, extreme gradient boost, random forest, and native Bayesian. In the test with real field data, four types of features, which include the manual features, the CNN features, and the combined features without and with selection, are compared with these different classifiers. As a result, it shows the combined features without selection can improve the identification ability of DAS compared with the recognition with only manual or deep features. The combined features with selection can further improve the computation efficiency and save up to 90% of time with a performance degradation of less than 1%.

© 2021 Optical Society of America

Pattern recognition in distributed fiber-optic acoustic sensor using an intensity and phase stacked convolutional neural network with data augmentation

Huan Wu, Bin Zhou, Kun Zhu, Chao Shang, Hwa-Yaw Tam, and Chao Lu
Opt. Express 29(3) 3269-3283 (2021)

Highly discriminative and adaptive feature extraction method based on NMF–MFCC for event recognition of Φ-OTDR

Yi Huang, Jingyi Dai, Wei Shen, Xiaofeng Chen, Chengyong Hu, Chuanlu Deng, Lin Chen, Xiaobei Zhang, Wei Jin, Jianming Tang, and Tingyun Wang
Appl. Opt. 62(35) 9326-9333 (2023)

Intensity and phase stacked analysis of a Φ-OTDR system using deep transfer learning and recurrent neural networks

Ceyhun Efe Kayan, Kivilcim Yuksel Aldogan, and Abdurrahman Gumus
Appl. Opt. 62(7) 1753-1764 (2023)

Machine learning methods for identification and classification of events in ϕ-OTDR systems: a review

Deus F. Kandamali, Xiaomin Cao, Manling Tian, Zhiyan Jin, Hui Dong, and Kuanglu Yu
Appl. Opt. 61(11) 2975-2997 (2022)

Intelligent water perimeter security event recognition based on NAM-MAE and distributed optic fiber acoustic sensing system

Mingyang Sun, Miao Yu, Haoran Wang, Kaiwen Song, Xinyu Guo, Songfeng Xue, Hongwei Zhang, Yanbin Shao, Hongliang Cui, Tianying Chang, and Tianyu Zhang
Opt. Express 31(22) 37058-37073 (2023)