基于目标检测的分布式光纤实时多源扰动识别技术研究

基于相位敏感光时域反射技术（Phase-Sensitive Optical Time-Domain Reflectometry，Φ-OTDR）的分布式光纤传感系统具有体积小，质量轻，灵敏度高，抗电磁干扰，长距离无缝监测等优点，在周界安防、局部放电监测、油气管道健康监测等场景中得到广泛应用。在这些应用场景下，对传感光纤上的扰动进行定位并对其类型分类是重要需求之一。随着Φ-OTDR系统应用场景的复杂化，以及用户对系统实时性的更高需求，传感系统也对扰动识别算法的实时性有了更高的要求。本文围绕Φ-OTDR分布式光纤传感系统中扰动识别方法展开研究，与目标检测算法结合，提出了基于Fast R-CNN和基于Faster R-CNN的两种扰动识别方法以及基于YOLOv3的实时扰动识别方法。本文完成的研究工作如下：

在理论研究Φ-OTDR分布式光纤传感原理和目标检测原理的基础上，搭建基于直接探测型Φ-OTDR的分布式光纤传感系统，针对周界安防场景进行实验及数据采集，研究传感数据处理方法，通过标注扰动信息构建传感数据集。

提出基于Fast R-CNN和基于Faster R-CNN的两种扰动识别方法，根据应用场景及数据特点对算法进行改进。在对其进行充分训练后，两种方法在测试集上分别达到了95.737%和96.832%的平均准确率，证明两种扰动识别方法均能有效对五种事件定位和分类。基于Fast R-CNN和基于Faster R-CNN的扰动识别方法在识别速度上分别达到了0.5 FPS（Frame Per Second）和6 FPS，具有良好的准确率和一定的实时性。

提出基于YOLOv3的实时扰动识别方法，根据应用场景、数据特点以及对定位准确度需求改进算法。使用训练集充分训练后，在测试集上得到96.141%的平均准确率，并且识别速度达到了22.83 FPS，远超前两种方法。随后针对视频化的传感信号进行实时扰动识别，并通过理论和工程应用分析，得出实际应用中的高实时性前提条件。通过评价指标对比及结果分析，证明提出的基于YOLOv3的实时扰动识别方法适用于对实时性高要求的场景，具有很高的实用价值。

The distributed optical fiber sensing system based on phase-sensitive optical time domain reflectometer (Φ-OTDR) has the advantages of small size, light weight, high sensitivity, anti-electromagnetic interference, and long-distance seamless monitoring, which is widely used in scenarios such as perimeter security, partial discharge monitoring, and pipeline health monitoring. In these application scenarios, locating and classifying the type of disturbance on the sensing fiber is one of the important requirements. As the application scenarios of Φ-OTDR system become more complex and the users demand higher real-time performance of the system, the sensing system also has higher requirements for the real-time performance of the disturbance recognition algorithm. In this paper, we propose two disturbance recognition methods based on Fast R-CNN and Faster R-CNN as well as a real-time disturbance recognition method based on YOLOv3, which are focused on Φ-OTDR distributed optical fiber sensing system. The research work accomplished in this paper is as follows.

Based on the theoretical study of Φ-OTDR distributed optical fiber sensing principle and object detection, we build a distributed optical fiber sensing system based on direct detection Φ-OTDR, conduct experiments and data acquisition for perimeter security scenarios, review sensing data processing methods, and construct sensing data sets by labeling disturbance information.

Two disturbance recognition methods based on Fast R-CNN and Faster R-CNN are proposed, and the algorithms are improved according to the application scenarios and data characteristics. After their adequate training, the average accuracy of 95.737% and 96.832% are achieved on the test set, which proves that both methods can effectively detect and classify the five disturbance events. The Fast R-CNN-based and Faster R-CNN-based disturbance recognition methods achieve 0.5 FPS and 6 FPS in recognition speed, with good accuracy and certain real-time performance.

The YOLOv3-based real-time disturbance recognition method is proposed, and the algorithm is improved according to the application scenario, data characteristics, and requirements for localization accuracy. After sufficient training using the training set, an average accuracy of 96.141% was obtained on the test set, and the recognition speed reached 22.83 FPS, far exceeding the previous two methods. Subsequently, real-time disturbance recognition is performed for sensing video, and the prerequisites for high real-time performance in practical applications are derived through theoretical and engineering application analysis. Through evaluation index comparison and result analysis, it is demonstrated that the proposed YOLOv3-based real-time disturbance recognition method is suitable for scenarios with high requirements for real-time performance and has high practical value.

[1] KERSEY A D, DAVIS M A, PATRICK H J, et al. Fiber grating sensors[J]. Journal of Lightwave Technology, 1997, 15(8): 1442-1463.
[2] JOUSSET P, REINSCH T, RYBERG T, et al. Dynamic strain determination using fibre-optic cables allows imaging of seismological and structural features[J]. Nature Communications, 2018, 9(1): 1-11.
[3] LINDSEY N J, DAWE T C, AJO-FRANKLIN J B. Illuminating seafloor faults and ocean dynamics with dark fiber distributed acoustic sensing[J]. Science, 2019, 366(6469): 1103-1107.
[4] WANG F, LIU Z, ZHOU X, et al. Oil and gas pipeline leakage recognition based on distributed vibration and temperature information fusion[J]. Results in Optics, 2021, 5: 100131.
[5] PENG F, DUAN N, RAO Y-J, et al. Real-time position and speed monitoring of trains using phase-sensitive OTDR[J]. IEEE Photonics Technology Letters, 2014, 26(20): 2055-2057.
[6] HUANG M-F, SALEMI M, CHEN Y, et al. First field trial of distributed fiber optical sensing and high-speed communication over an operational telecom network[J]. Journal of Lightwave Technology, 2019, 38(1): 75-81.
[7] MIN R, LIU Z, PEREIRA L, et al. Optical fiber sensing for marine environment and marine structural health monitoring: A review[J]. Optics & Laser Technology, 2021, 140: 107082.
[8] TEJEDOR J, MACIAS-GUARASA J, MARTINS H F, et al. Real field deployment of a smart fiber-optic surveillance system for pipeline integrity threat detection: Architectural issues and blind field test results[J]. Journal of Lightwave Technology, 2017, 36(4): 1052-1062.
[9] ROHWETTER P, EISERMANN R, KREBBER K. Distributed acoustic sensing: Towards partial discharge monitoring[C]//24th International Conference on Optical Fibre Sensors. SPIE, 2015, 9634: 125-128.
[10] CHEN Z, ZHANG L, LIU H, et al. 3D printing technique-improved phase-sensitive OTDR for breakdown discharge detection of gas-insulated switchgear[J]. Sensors, 2020, 20(4): 1045.
[11] SHAO L, LIU S, BANDYOPADHYAY S, et al. Data-driven distributed optical vibration sensors: A review[J]. IEEE Sensors Journal, 2020, 20(12): 6224-6239.
[12] TAYLOR H F , LEE C E . Apparatus and method for fiber optic intrusion sensing: US, US5194847 A[P]. 1993.
[13] PARK J, LEE W, TAYLOR H F. Fiber optic intrusion sensor with the configuration of an optical time-domain reflectometer using coherent interference of Rayleigh backscattering[C]//Optical and Fiber Optic Sensor Systems. SPIE, 1998, 3555: 49-56.
[14] JUAREZ J C, MAIER E W, CHOI K N, et al. Distributed fiber-optic intrusion sensor system[J]. Journal of Lightwave Technology, 2005, 23(6): 2081-2087.
[15] JUAREZ J C, TAYLOR H F. Field test of a distributed fiber-optic intrusion sensor system for long perimeters[J]. Applied Optics, 2007, 46(11): 1968-1971.
[16] LU Y, ZHU T, CHEN L, et al. Distributed vibration sensor based on coherent detection of phase-OTDR[J]. Journal of Lightwave Technology, 2010, 28(22): 3243-3249.
[17] LIU S, YU F, HONG R, et al. Advances in phase-sensitive optical time domain reflectometry[J]. Opto-Electronic Advances, 2022: 200078-200071-200078-200028.
[18] ZHU H, PAN C, SUN X. Vibration pattern recognition and classification in OTDR based distributed optical-fiber vibration sensing system[J]. Proceedings of SPIE - The International Society for Optical Engineering, 2013, 9062.
[19] SUN Q, FENG H, YAN X, et al. Recognition of a phase-sensitivity OTDR sensing system based on morphologic feature extraction[J]. Sensors, 2015, 15(7): 15179-15197.
[20] WU H, XIAO S, LI X, et al. Separation and determination of the disturbing signals in phase-sensitive optical time domain reflectometry (Φ-OTDR)[J]. Journal of Lightwave Technology, 2015, 33(15): 3156-3162.
[21] PAPP A, WIESMEYR C, LITZENBERGER M, et al. A real-time algorithm for train position monitoring using optical time-domain reflectometry[C]//2016 IEEE International Conference on Intelligent Rail Transportation (ICIRT). IEEE, 2016: 89-93.
[22] 张俊楠, 娄淑琴, 梁生. 基于 SVM 算法的 Φ-OTDR 分布式光纤扰动传感系统模式识别研究 [J]. 红外与激光工程 , 2017, 46(4): 422003-0422003 (0422007).
[23] XU C, GUAN J, BAO M, et al. Pattern recognition based on enhanced multifeature parameters for vibration events in Φ-OTDR distributed optical fiber sensing system[J]. Microwave and Optical Technology Letters, 2017, 59(12): 3134-3141.
[24] XU C, GUAN J, BAO M, et al. Pattern recognition based on time-frequency analysis and convolutional neural networks for vibrational events in Φ-OTDR[J]. Optical Engineering, 2018, 57(1): 016103.
[25] WANG X, LIU Y, LIANG S, et al. Event identification based on random forest classifier for Φ-OTDR fiber-optic distributed disturbance sensor[J]. Infrared Physics & Technology, 2019, 97: 319-325.
[26] SHI Y, WANG Y, ZHAO L, et al. An event recognition method for Φ-OTDR sensing system based on deep learning[J]. Sensors, 2019, 19(15): 3421.
[27] WU H, CHEN J, LIU X, et al. One-dimensional CNN-based intelligent recognition of vibrations in pipeline monitoring with DAS[J]. Journal of Lightwave Technology, 2019, 37(17): 4359-4366.
[28] WANG Z, ZHENG H, LI L, et al. Practical multi-class event classification approach for distributed vibration sensing using deep dual path network[J]. Optics Express, 2019, 27(17): 23682-23692.
[29] FOUDA B M T, YANG B, HAN D, et al. Pattern recognition of optical fiber vibration signal of the submarine cable for its safety[J]. IEEE Sensors Journal, 2020, 21(5): 6510-6519.
[30] LYU C, HUO Z, CHENG X, et al. Distributed optical fiber sensing intrusion pattern recognition based on GAF and CNN[J]. Journal of Lightwave Technology, 2020, 38(15): 4174-4182.
[31] WU H, ZHOU B, ZHU K, et al. Pattern recognition in distributed fiber-optic acoustic sensor using an intensity and phase stacked convolutional neural network with data augmentation[J]. Optics Express, 2021, 29(3): 3269-3283.
[32] JIANG F, LI H, ZHANG Z, et al. Localization and discrimination of the perturbation signals in fiber distributed acoustic sensing systems using spatial average kurtosis[J]. Sensors, 2018, 18(9): 2839.
[33] CAO C, FAN X, LIU Q, et al. Practical pattern recognition system for distributed optical fiber intrusion monitoring system based on phase-sensitive coherent OTDR[C]//Asia Communications and Photonics Conference. Optical Society of America, 2015: ASu2A. 145.
[34] TEJEDOR J, MARTINS H F, PIOTE D, et al. Toward prevention of pipeline integrity threats using a smart fiber-optic surveillance system[J]. Journal of Lightwave Technology, 2016, 34(19): 4445-4453.
[35] TEJEDOR J, MACIAS-GUARASA J, MARTINS H F, et al. A novel fiber optic based surveillance system for prevention of pipeline integrity threats[J]. Sensors, 2017, 17(2): 355.
[36] WU H, LI X, LI H, et al. An effective signal separation and extraction method using multi-scale wavelet decomposition for phase-sensitive OTDR system[C]//Sixth International Symposium on Precision Mechanical Measurements. International Society for Optics and Photonics, 2013, 8916: 89160Z.
[37] WU H, QIAN Y, ZHANG W, et al. Feature extraction and identification in distributed optical-fiber vibration sensing system for oil pipeline safety monitoring[J]. Photonic Sensors, 2017, 7(4): 305-310.
[38] LI Q, ZHANG C, LI C. Fiber-optic distributed sensor based on phase-sensitive OTDR and wavelet packet transform for multiple disturbances location[J]. Optik, 2014, 125(24): 7235-7238.
[39] HUI X, ZHENG S, ZHOU J, et al. Hilbert–huang transform time-frequency analysis in Φ-OTDR distributed sensor[J]. IEEE Photonics Technology Letters, 2014, 26(23): 2403-2406.
[40] WANG B, PI S, SUN Q, et al. Improved wavelet packet classification algorithm for vibrational intrusions in distributed fiber-optic monitoring systems[J]. Optical Engineering, 2015, 54(5): 055104.
[41] AKTAS M, AKGUN T, DEMIRCIN M U, et al. Deep learning based multi-threat classification for phase-OTDR fiber optic distributed acoustic sensing applications[C]//Fiber Optic Sensors and Applications XIV. International Society for Optics and Photonics, 2017, 10208: 102080G.
[42] WEN H, PENG Z, JIAN J, et al. Artificial intelligent pattern recognition for optical fiber distributed acoustic sensing systems based on phase OTDR[C]//2018 Asia Communications and Photonics Conference (ACP). IEEE, 2018: 1-4.
[43] MARIE T F B, HAN D, AN B, et al. A research on fiber-optic vibration pattern recognition based on time-frequency characteristics[J]. Advances in Mechanical Engineering, 2018, 10(12): 1687814018813468.
[44] CHANG C-C, LIN C-J. LIBSVM: a library for support vector machines[J]. ACM Transactions on Intelligent Systems and Technology (TIST), 2011, 2(3): 1-27.
[45] 张旭苹. 全分布式光纤传感技术[M]. 北京: 科学出版社, 2013: 27-28
[46] BARNOSKI M, JENSEN S. Fiber waveguides: a novel technique for investigating attenuation characteristics[J]. Applied Optics, 1976, 15(9): 2112-2115.
[47] BARNOSKI M K, ROURKE M D, JENSEN S, et al. Optical time domain reflectometer[J]. Applied Optics, 1977, 16(9): 2375-2379.
[48] HEALEY P, MALYON D. OTDR in single-mode fibre at 1.5 μm using heterodyne detection[J]. Electronics Letters, 1982, 18(20): 862-863.
[49] EICKHOFF W, ULRICH R. Optical frequency domain reflectometry in single‐mode fiber[J]. Applied Physics Letters, 1981, 39(9): 693-695.
[50] ROGERS A. Polarisation optical time domain reflectometry[J]. Electronics Letters, 1980, 16(13): 489-490.
[51] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: Unified, real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2016: 779-788.
[52] UIJLINGS J R, VAN DE SANDE K E, GEVERS T, et al. Selective search for object recognition[J]. International Journal of Computer Vision, 2013, 104(2): 154-171.
[53] ZITNICK C L, DOLLáR P. Edge boxes: Locating object proposals from edges[C]//European Conference on Computer Vision. Springer, Cham, 2014: 391-405.
[54] CHOI J-Y, SUNG K-S, YANG Y-K. Multiple vehicles detection and tracking based on scale-invariant feature transform[C]//2007 IEEE Intelligent Transportation Systems Conference. IEEE, 2007: 528-533.
[55] BAY H, TUYTELAARS T, GOOL L V. Surf: Speeded up robust features[C]//European Conference on Computer Vision. Springer, Berlin, Heidelberg, 2006: 404-417.
[56] CORVEE E, BREMOND F. Body parts detection for people tracking using trees of histogram of oriented gradient descriptors[C]//2010 7th IEEE International Conference on Advanced Video and Signal Based Surveillance. IEEE, 2010: 469-475.
[57] SUYKENS J A, VANDEWALLE J. Least squares support vector machine classifiers[J]. Neural processing letters, 1999, 9(3): 293-300.
[58] RäTSCH G, ONODA T, MüLLER K-R. Soft margins for AdaBoost[J]. Machine learning, 2001, 42(3): 287-320.
[59] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. Imagenet classification with deep convolutional neural networks[J]. Advances in Neural Information Processing Systems, 2012, 25
[60] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2014: 580-587.
[61] GIRSHICK R. Fast r-cnn[C]//Proceedings of the IEEE International Conference on Computer Vision. 2015: 1440-1448.
[62] REN S, HE K, GIRSHICK R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149.
[63] LIU W, ANGUELOV D, ERHAN D, et al. Ssd: Single shot multibox detector[C]//European Conference on Computer Vision. Springer, Cham, 2016: 21-37.
[64] WANG Z, LI J, FAN M, et al. Phase-sensitive optical time-domain reflectometry with Brillouin amplification[J]. Optics Letters, 2014, 39(15): 4313-4316.
[65] HE H, LUO B, ZOU X, et al. Enhanced phase-sensitive OTDR system with pulse width modulation Brillouin amplification[J]. Optics Express, 2018, 26(18): 23714-23727.
[66] MARTINS H F, MARTíN-LóPEZ S, CORREDERA P, et al. Phase-sensitive optical time domain reflectometer assisted by first-order Raman amplification for distributed vibration sensing over>100 km[J]. Journal of Lightwave Technology, 2014, 32(8): 1510-1518.
[67] PENG F, WU H, JIA X-H, et al. Ultra-long high-sensitivity Φ-OTDR for high spatial resolution intrusion detection of pipelines[J]. Optics Express, 2014, 22(11): 13804-13810.
[68] WANG Z, ZENG J, LI J, et al. Ultra-long phase-sensitive OTDR with hybrid distributed amplification[J]. Optics Letters, 2014, 39(20): 5866-5869.
[69] ÖLçER İ, ÖNCü A. Adaptive temporal matched filtering for noise suppression in fiber optic distributed acoustic sensing[J]. Sensors, 2017, 17(6): 1288.
[70] QIN Z, CHEN L, BAO X. Wavelet denoising method for improving detection performance of distributed vibration sensor[J]. IEEE Photonics Technology Letters, 2012, 24(7): 542-544.
[71] QIN Z, CHEN L, BAO X. Continuous wavelet transform for non-stationary vibration detection with phase-OTDR[J]. Optics Express, 2012, 20(18): 20459-20465.
[72] QIN Z, CHEN H, CHANG J. Detection performance improvement of distributed vibration sensor based on curvelet denoising method[J]. Sensors, 2017, 17(6): 1380.
[73] QIN Z, CHEN H, CHANG J. Signal-to-noise ratio enhancement based on empirical mode decomposition in phase-sensitive optical time domain reflectometry systems[J]. Sensors, 2017, 17(8): 1870.
[74] ZHU T, XIAO X, HE Q, et al. Enhancement of SNR and spatial resolution in Φ-OTDR system by using two-dimensional edge detection method[J]. Journal of Lightwave Technology, 2013, 31(17): 2851-2856.
[75] HE H, SHAO L, LI H, et al. SNR enhancement in phase-sensitive OTDR with adaptive 2-D bilateral filtering algorithm[J]. IEEE Photonics Journal, 2017, 9(3): 1-10.
[76] SOTO M A, RAMIREZ J A, THéVENAZ L. Intensifying the response of distributed optical fibre sensors using 2D and 3D image restoration[J]. Nature Communications, 2016, 7(1): 1-11.
[77] MURPHY K P. Machine learning: a probabilistic perspective[M]. MIT press, 2012: 180-184
[78] HE K, ZHANG X, REN S, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9): 1904-1916.
[79] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. arXiv preprint arXiv:1409.1556, 2014
[80] REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017: 7263-7271.
[81] REDMON J, FARHADI A. Yolov3: An incremental improvement[J]. arXiv preprint arXiv:1804.02767, 2018
[82] LIN T-Y, DOLLáR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE conference on Computer Vision and Pattern Recognition. 2017: 2117-2125.
[83] SHI Y, LI Y, ZHANG Y, et al. An easy access method for event recognition of Φ-OTDR sensing system based on transfer learning[J]. Journal of Lightwave Technology, 2021, 39(13): 4548-4555.
[84] RUSSAKOVSKY O, DENG J, SU H, et al. Imagenet large scale visual recognition challenge[J]. International Journal of Computer Vision, 2015, 115(3): 211-252.
[85] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Visionand Pattern Recognition. 2016: 770-778.f