小波多尺度信息综合的行波波头检测算法研究  

Traveling Wave Head Detection Algorithm Based on Wavelet Multiscale Information Fusion

作　　者：郑楚韬 孔祥轩 关家华 谭家琪 陆凯烨 王伟冠 游金梁 林刚[2] 张朕 ZHENG Chutao;KONG Xiangxuan;GUAN Jiahua;TAN Jiaqi;LU Kaiye;WANG Weiguan;YOU Jinliang;LIN Gang;ZHANG Zhen(Guangdong Foshan Power Supply Bureau,Foshan 528200, China;School of Electrical and Engineering, Wuhan University,Wuhan 430072, China)

机构地区：[1]广东电网有限责任公司佛山供电局,广东佛山528200 [2]武汉大学电气与自动化学院,湖北武汉430072

出　　处：《智慧电力》2019年第5期97-102,共6页Smart Power

基　　金：国家自然科学基金资助项目(51777142)~~

摘　　要：提出了一种小波变换多尺度信息综合的行波波头检测算法,首先利用小波变换的多尺度分析对故障行波信号进行处理,得到各个尺度下的小波系数模极大值,然后将模极大值依次标定在平行坐标系下并形成灰度图片,最后利用卷积神经网络搭建波头检测模型。本算法利用平行坐标系巧妙的将小波变换与CNN相联系,结合了小波变换多尺度下的信息,为行波波头的检测提供了一种新思路。The key to traveling wave protection and ranging technology lies in the detection of fault wave heads, which is also the focus of research. Wavelet multi-scale information fusion based traveling wave detection algorithm is proposed. Firstly, the wavelet signal is processed by multi-scale analysis of wavelet transform to get the modular maxima at each scale. Then the modular maxima are calibrated in parallel coordinate system and gray scale images are formed. Finally, a wave head detection model is constructed using convolutional neural network. This algorithm uses the parallel coordinate system to intelligently associate the wavelet transform with CNN,which provides a new idea for the detection of traveling wave heads.

关 键 词：行波波头 卷积神经网络 小波变换 平行坐标系 

分 类 号：TM771[电气工程—电力系统及自动化]