微振动传感与声振特征识别的输电线路环境异变预警研究  

Research on Early Warning of Environmental Change of Transmission Line Based on Micro-vibration Sensing and Acoustic Vibration Feature Recognition

作　　者：何冰 孟夏卿 顾俊杰 俞杰 李伟[3] HE Bing;MENG Xiaqing;GU Junjie;YU Jie;LI Wei(Maintenance Company,State Grid Shanghai Electric Power Company,Shanghai 200063,China;Hylight Technology Co.,Ltd.,Shanghai 201800,China;Shanghai Institute of Microsystem and Information Technology,Chinese Academy of Sciences,National Key Laboratory for Information Functional Materials,Shanghai 200050,China)

机构地区：[1]国网上海市电力公司检修公司,上海200063 [2]中光华研电子科技有限公司,上海201800 [3]中国科学院上海微系统与信息技术研究所信息功能材料国家重点实验室,上海200050

出　　处：《电力信息与通信技术》2020年第9期57-63,共7页Electric Power Information and Communication Technology

基　　金：国家电网上海市电力公司科技项目资助“基于NB-IOT的智能铁塔运检关键技术研究与应用”(52095019001Y)。

摘　　要：输电线路常常遭受施工机械破坏导致停电事故,而传统的人工巡检和视频监控不具有实时性,光纤传感“后知后觉”,不能起到及时预警的作用。针对上述情况,文章通过研究声信号的特性及其在传播过程中的衰减特性,分析工程机械声信号的产生机理。并基于微振动传感器开发振动采集装置,现场采集各类工程机械不同条件下的声信号数据,分析声信号的振动特征,通过预处理和梅尔频率倒谱系数(Mel Frequency Cepstrum Coefficient,MFCC)特征提取,构建模型数据库,利用对数似然比的评价指标进行被测声信号识别。实际测试结果表明,典型工程机械的识别率达到93%以上,验证了声信号振动特征的准确性和识别算法的有效性,实现了输电线路环境异变监测和预警。Transmission lines are often damaged by construction machinery leading to power outages,while traditional manual inspections and video surveillance do not have real-time performance.Fiber optic sensors are"post-perceived"and cannot serve as a timely warning.In view of the above situation,this paper analyzes the generation mechanism of acoustic signals of construction machinery by studying the characteristics of acoustic signals and their attenuation characteristics during propagation.Based on the micro vibration sensor,a vibration acquisition device is developed to collect sound signal data of various construction machinery under different conditions on the spot,by analyzing the vibration characteristics of the sound signal,a model database through preprocessing and MFCC feature extraction is built,and a vibration feature recognition algorithm is developed.The actual test results show that the recognition rate of typical construction machinery reaches more than 93%,which verifies the accuracy of the acoustic signal vibration characteristics and the effectiveness of the recognition algorithm.It realizes the monitoring and early warning of transmission line environmental changes.

关 键 词：微振动传感器 特征识别 输电线路 

分 类 号：TN912.1[电子电信—通信与信息系统]