Research On Pipeline Health Status Monitoring Based On Distributed Optical Fiber Sensing

Nowadays,various transmission pipelines have gradually become important infrastructures in countries all over the world,and many important resources are transported through pipelines.At present,there are millions of kilometers of pipelines in the world,and the construction of a large number of pipeline networks has brought serious pipeline maintenance problems.During the year-round operation of pipelines,third-party invasion and damage,pipeline corrosion,blockage and leakage will cause pipeline failure.Therefore,how to grasp the operation status of the pipeline in real time and ensure its normal operation is of great research significance.In response to the growing demand for long-distance pipeline health status monitoring,this paper proposes a distributed optical fiber vibration sensing technology based on phase-sensitive optical time domain reflectometry(Phase-sensitive Optical Time Domain Reflectometry,φ-OTDR).Provide early warning and alarm for abnormal conditions such as blockage and corrosion during pipeline operation and external abnormal disturbances,and realize real-time and comprehensive monitoring of pipeline health status.The specific work content is as follows:1.Simplified the modeling of the common internal abnormal state of the pipeline,combined with the ANSYS Workbench two-way fluid-solid coupling module to simulate the pipeline operation state,and analyzed five types of pipe conditions of two degrees of blockage,two degrees of defects,and normal pipelines.Research including:(1)The study analyzed the differences in fluid velocity,pressure,turbulence kinetic energy and the form of action on the pipe wall under different pipe conditions.(2)The frequency-domain signal characteristics of pipeline flow-induced vibration are studied and analyzed,and the change law of frequencydomain information under different pipeline conditions is summarized through the spectrogram.(3)Through the wavelet packet decomposition,the energy variation law of the vibration of the pipe wall in different frequency bands under different pipe conditions is proved,which provides a theoretical basis for the subsequent construction of the experimental platform,signal acquisition and data processing and identification.From a qualitative point of view,the feasibility of distinguishing and identifying different abnormal states of pipelines through frequency domain features is demonstrated.2.Build a pipeline abnormal state monitoring platform,simulate the actual pipeline working state,make a physical model of pipeline abnormality,and connect it to the working pipeline to simulate different types of abnormal states that occur during pipeline operation.The distributed optical fiber vibration sensing system based on φ-OTDR is used to collect a large amount of data on the vibration information of pipelines with different abnormal pipeline conditions.3.Combined with the conclusions obtained in the pipeline simulation,the collected vibration phase information is decomposed into multi-layer wavelet packets,and the wavelet energy of different nodes is calculated.Analyze the differences in the energy distribution of each frequency band under different types of abnormal states in the experiment,and use it as a signal feature to train the BP neural network.Finally,the model obtained through training classifies and recognizes the collected data,and evaluates and analyzes the recognition results based on indicators such as accuracy rate,precision rate,recall rate and F1-score,and obtains a good recognition effect.