Zhen Yang, Hao Feng. Oil Pipeline Intrusion Monitoring Based on Deep Learning of Φ-OTDR[J]. Laser & Optoelectronics Progress, 2022, 59(8): 0806001
- Laser & Optoelectronics Progress
- Vol. 59, Issue 8, 0806001 (2022)
Fig. 1. Structure of Φ-OTDR system
Fig. 2. Space-time diagram
Fig. 3. Schematic diagram of pipeline and optical fiber laying
Fig. 4. Experimental setup and environment
Fig. 5. Space-time diagram before and after preprocessing. (a) Original space-time diagram; (b) normalization; (c) bandpass filtering
Fig. 6. Data augmentation operations. (a) Origin picture; (b) displacement; (c) scale; (d) color jittering
Fig. 7. Space-time diagrams of three types of events. (a) Jumping; (b) beating; (c) digging
Fig. 8. Training and recognition speed of YOLOv3 with different structural feature extraction networks
Fig. 9. Precision and recall of YOLOv3 with different structural feature extraction networks
Fig. 10. Structure of overall network
Fig. 11. Improved feature extraction network
Fig. 12. Network detection results
Fig. 13. Loss curve of training
Fig. 14. Precision curve and recall curve of training
|
Table 1. Influence of different data augmentation methods on network effect
|
Table 2. Sample type and quantity
|
Table 3. Recall and accuracy of different types of events
Set citation alerts for the article
Please enter your email address
© Copyright 2018-2021 | Chinese Laser Press. All Rights Reserved 沪ICP备15018463号-20