Two-Stage Intrusion Events Recognition for Vibration Signals from Distributed Optical Fiber Sensors
Zhuoling Lyu,Chengyuan Zhu,Yanyun Pu,Zuan Chen,Kaixiang Yang,Qinmin Yang
DOI: https://doi.org/10.1109/tim.2023.3336173
IF: 5.6
2024-01-01
IEEE Transactions on Instrumentation and Measurement
Abstract:Smart pipeline systems (SPSs) based on phase-sensitive optical time-domain reflectometry (phi-OTDR) distributed optical fiber sensors (DOFSs) are widely used to recognize and locate third-party events that may damage long-distance pipelines. However, the geological surroundings and infrastructures along the pipeline are highly complex, encompassing densely populated highways, mountains, railways, and so on, and thus, the vibration intensities also vary greatly. Moreover, the system in real sites always entails high computation costs, leading to lengthy delays that cannot meet real-time demands. In this article, a data-driven two-stage early warning strategy is designed to monitor the safety of oil and gas pipelines. Specifically, the quality of classification model can be greatly enhanced by identifying and excising redundant values through a preselection mechanism. In addition, the study proposes a model based on cascade forest to recognize events. Extensive comparative experiments on real-world datasets demonstrate the superiority of the proposed methods in terms of both effectiveness and efficiency.