基于EMD-HHT可定位长输天然气管道第三方破坏事件监测技术

为适应目前管道安全监测需要，满足对扰动信号分类监测的实际需求，提出1种基于希尔伯变换和经验模态分解(EMD-HHT)的信号特征提取技术，利用基于φ-OTDR分布式光纤传感系统采集振动信号，通过EMD+HHT区分算法对管道沿线振动事件进行分解并提取6个典型特征向量，各特征事件数据经过EMD后选取IMF3为最终提取特征向量的原始数据，BP神经元网络可有效识别机械破坏、敲击破坏、车辆经过、人工挖掘、动力干扰5种事件。研究结果表明：在长输管道信号识别中，BP神经网络对5类事件平均识别率高达98.6%,该技术分类识别5类事件扰动信号，能够达到较高准确性，并且误报率平均在1.3%,能较好满足现场安全实时监测需求。研究结果对长输油气管道附近第三方破坏扰动信号分类监测具有一定参考意义。

Monitoring technology of third-party damage events in long-distance natural gas pipeline located by EMD-HHT

In order to meet the current needs of pipeline safety monitoring,and solve the actual needs of classification monitoring on disturbance signals,a signal feature extraction technique based on Hilberg transform and empirical mode decomposition (EMD-HHT) was proposed,and the vibration signals were collected by a distributed optical fiber sensing system based on φ-OTDR.The vibration events along the pipeline were decomposed by EMD+HHT distinguishing algorithm,and six typical feature vectors were extracted.After EMD,IMF3 is selected as the original data of feature vector extraction.The BP neural network could effectively identify five kinds of interference events,such as mechanical damage,knock damage,vehicle passing,artificial excavation and dynamic interference.The results showed that in the recognition of long-distance pipeline signals,the average recognition rate of BP neural network for 5 types of events was as high as 98.6%.The classification and recognition of 5 types of event disturbance signals could achieve high accuracy,and the average probability of false report was 1.3%,which could better meet the needs of field safety real-time monitoring.The research results have reference significance for the classification monitoring of disturbance signals.