基于EMD的GIS典型局部放电超高频信号的分形特征提取方法  

Fractal Dimension Research of the Partial Discharge UHF Signal in GIS Based on EMD

作　　者：李化 程昌奎 陈娇 干建伟 胡琳 李馨博 

机构地区：[1]成都供电公司,成都610041

出　　处：《高压电器》2014年第6期104-110,共7页High Voltage Apparatus

摘　　要：笔者提出了一种基于经验模式分解的GIS局放超高频信号分形特征提取与识别方法。该方法通过对局部放电超高频信号进行经验模式分解得到多个固有模态,通过计算固有模态的关联维数作为特征量进行模糊聚类识别。文中设计了4种典型局放模型获取局放超高频信号样本进行验证该方法,结果表明识别正确率最低达到98%。This paper presents a new characteristic quantity extracting method for PD UHF signal based on EMD analysis and fractal dimension. The EMD was used to decomposing the PD UHF signals, it can be obtain the IMF components, the fractal dimension of the IMF is calculated by the G-P method. Four typical GIS PD models are used to generate discharge in experiment, and the PD UHF signals are used as the calculating samplings. The results show that the method presented in the paper have high effectiveness, the lowest recognition rate can be reached at 98%.

关 键 词：GIS 局部放电 超高频信号 经验模式分解 关联维数 

分 类 号：TM643[电气工程—电力系统及自动化]