基于FSWT——SSAEs的配电网内部过电压自动提取与分类识别  

Automatically Extracting and Classification Recognition Internal Overvoltage Measured in Distribution Networks Based on FSWT-SSAEs

作　　者：陈钦柱 张涵 赵海龙 袁涛[2] 姚冬 司马文霞[2] CHEN Qinzhu;ZHANG Han;ZHAO Hailong;YUAN Tao;YAO Dong;SIMA Wenxia(Key Laboratory of Physical and Chemical Analysis for Electric Power of Hainan Province,Haikou 570311,China;State Key Laboratory of Power Transmission Equipment&System Security and New Technology,Chongqing University,Chongqing 400030,China)

机构地区：[1]海南省电网理化分析重点实验室,海口570311 [2]重庆大学输配电装备及系统安全与新技术国家重点实验室,重庆400030

出　　处：《高压电器》2020年第7期166-172,共7页High Voltage Apparatus

基　　金：国家重点研发计划(2017YFB0902701);海南电网有限责任公司科技项目(073000KK52170006);国家自然科学基金(51837002)。

摘　　要：过电压是造成电网绝缘损坏的主要原因,对电气设备绝缘可靠性、系统绝缘配合、继电保护以及运行控制均产生重要影响。研究配电网过电压的特征提取与分类识别对于电网运行事故溯源以及设备绝缘风险评估等均具有难以替代的现实意义。文中基于频率切片小波变换(FSWT)时频分析方法构建过电压时频分布九宫图,完成实测过电压整体与细节信息的完全提取;改进多层稀疏自编码算法(SSAEs),实现实测过电压特征的自动提取与分类识别;分析改进多层稀疏自编码网络中关键参数(卷积块大小、卷积特征数量以及稀疏性参数)的影响,确定最优参数,实现最佳分类识别效果。结果表明,过电压时频分布九宫图与改进多层稀疏自编码算法相结合能够高效的自动提取和分类实测过电压波形,分类精度良好。The overvoltage is the main cause of insulation damage in power grid,it has an important impact on the insulation reliability of electrical equipment,system insulation coordination,relay protection and operation control.It is of great practical significance to study the feature extraction and classification of the measured overvoltage in the distribution network for tracing the source of power grid operation accidents and risk assessment of equipment insulation,etc.This paper constructed the band of time-frequency distribution of overvoltage in Lo Shu Square based on FSWT and completed the overall and detail information of overvoltage extraction.The measured overvoltage feature automatically extraction and classification is achieved based on modified SSAEs.The influence of key parameters,namely,the size of convolutional patches,the number of convolutional maps and sparsity parameter in modified SSAEs are analyzed respectively,and the best optimization parameters are determined,the best classification and recognition effect is achieved.The results show that the combination of the Lo Shu Square of the time-frequency distribution of overvoltage and the modified SSAEs algorithm can automatically extract and classify the measured overvoltage waveform with good classification accuracy.

关 键 词：实测过电压 时频分布图 九宫图 多层稀疏自编码 分类识别 

分 类 号：TM933.2[电气工程—电力电子与电力传动]