Distributed fiber optic sensing signal recognition based on class-incremental learning

• Published in: Optical fiber technology 2024-10, Vol.87, p.103940, Article 103940
• Main Authors: Liu, Zhaoying, Zhang, Faxiang, Sun, Zhihui, Jiang, Shaodong, Duan, Zhenhui
• Format: Article
• Language: English
• Subjects: Distributed fiber optics; Incremental learning; Machine learning; φ-OTDR
• ISSN: 1068-5200
• DOI: 10.1016/j.yofte.2024.103940

•A class-incremental learning scheme for distributed fiber optic intrusion signal recognition.•Utilized a new recognition model combining improved ECA with the ConvNeXt network.•Implemented the optimized LwM algorithm to alleviate forgetting for old classes.•Added a Linear Correction Unit to correct the deviation for new classes. Distributed fiber optic sensing (DFOS) based on phase-sensitive optical time-domain reflectance (φ-OTDR) technology has outstanding performance in pipeline safety monitoring and perimeter security detection. Accurate identification of new events remains challenging due to environmental variability and emerging forms of intrusions. In order to solve the problem of failing to accurately identify new events due to the inability to obtain all samples at once in real-time monitoring, this paper proposes an incremental learning network framework for distributed fiber-optic sensing signal recognition. This framework integrates an optimized Learning without Memorizing (LwM) algorithm with an improved ConvNeXt network for dynamic training of new events. An improved Efficient Channel Attention (HECA) is incorporated to thoroughly extract the spatio-temporal features of the intrusion signals collected by the DFOS. The forgetting problem is mitigated during incremental learning using knowledge distillation and optimized Gradient Weighted Class Activation Mapping to generate an attention map. A linear correction layer is added after the output layer to correct the bias towards new classes by rebalancing the information between new and old classes. Experimental comparisons show that the recognition rate for 10 different intrusion signals exceeds 93 %, while the forgetting rate is reduced from a peak of 41.44 % to 5.25 %. The time required to process and train incremental learning for 1000 samples in real time on an edge device (NVIDIA 3050 GPU) is approximately 1060 s, its ability to demonstrating its suitability for deployment in resource-constrained.