电力通信网络线路常因沿途的工程机械施工,人为外力破坏等行为造成毫无预警的危害和损坏。为了实现对线路外破隐患高效预警,需要在现有预警定位的基础上,加上对碰撞敲击、管孔侵入、机械施工和开盖报警等行为的识别。将分布式光纤传感技术、梅尔频率倒谱系数和径向基神经网络相结合,提出一种可应用于光缆外破在线监控的模式识别方案。测试的总体识别率达97.78%,现场模拟的事件报警准确率高。该方案可广泛应用于长距离管线检测领域。

Power line communication network always suffer from the engineering construction and artificial damage without warning.In order to realize the effective early warning to the hidden danger of line,we need to add some pattern recognition scheme on the basis of the existing early-alert,which can distinguish different vibration signal such as knock,pipeline intrusion,mechanical construction and manhole covers opening.In this paper,the distributed optical fiber sensing technology,Mel frequency cepstral coefficient and radial basis function neural network are used in this optical cable online monitoring pattern recognition scheme.The overall recognition rate of the test sample is 97.78%.The accuracy of the incident alarm is remarkable in the scene simulation.The scheme can be widely used in the area of long-distance pipe detection.