Long-Distance Pipeline Intrusion Warning Based on Environment Embedding from Distributed Optical Fiber Sensing

Chengyuan Zhu,Yanyun Pu,Zhuoling Lyu,Jiren Qian,Kaixiang Yang

DOI: https://doi.org/10.1007/978-3-031-44204-9_22

2023-01-01

Abstract:Pipeline is one of the most important transportation modes of oil and gas. However, it is different from the single-scene environments in fields such as security and high-speed railway, as the environment along the length of long-distance pipelines is complex and variable, including plains, mountains, and lakes. Different geological conditions have different effects on the transmission of vibrations, which poses great challenges to the external perception of distributed optic fiber sensing (DOFS) systems. At the same time, the similarity of different external activity signal characteristics also greatly affects the recognition performance of DOFS systems. Mechanical construction activities along the pipeline have posed a serious threat to the safety of pipelines. In this paper, an intrusion warning ensemble model based on environmental embedding is developed on the latest novel distributed optic fiber hardware system (f-OTDR), named Surroundings-Embedding Ensemble Learning (SEEL). Environmental embedding technology captures the fine-grained differences in spatial environment between different defense areas. At the same time, ensemble learning technology reduces the negative impact of random noise and environmental interference. A large amount of data collected in the actual environment is used for comparison experiments. The results show that this model can achieve more accurate intrusion detection accuracy and has a wider practical application range by effectively fusing environmental information. In addition, the effectiveness of the new component designed is verified through ablation experiments.