用谐波小波包变换法提取GIS局部放电信号多尺度特征参数

发布时间：2018-05-15 16:52

本文选题：谐波小波包 + 特征提取　；参考：《电工技术学报》2015年03期

【摘要】：超高频(UHF)法在GIS局部放电(PD)检测中已得到了广泛应用,UHF PD信号的特征提取对准确识别GIS内部绝缘缺陷类型和指导检修工作具有重要意义,但目前仍然缺乏有效的特征提取方法。为此,本文利用谐波小波具有严格盒形频谱的优点,提出一种提取UHF PD特征信息的谐波小波包变换(HWPT)方法,对实验室获取的4种典型放电模型产生的UHF PD信号,采用HWPT进行多尺度分解,以克服实小波包分解子带间存在频谱混叠和能量泄漏的缺陷,利用UHF PD信号在不同尺度能量和复杂度的差异,提取多尺度能量和多尺度样本熵参数作为模式识别的特征量,更加精确地描述了UHF PD信号的时频域信息。最后利用支持向量机分类识别的结果表明,该方法可以取得比实小波包更好的识别效果,多尺度能量和多尺度样本熵特征参数均能有效识别4种绝缘缺陷。
[Abstract]:UHF (UHF) method has been widely used in the detection of GIS partial discharge (PD) signals. The feature extraction of UHF PD signals is of great significance for accurately identifying the types of internal insulation defects in GIS and guiding the repair work. However, there is still a lack of effective feature extraction methods at present. In this paper, a harmonic wavelet packet transform (HWPT) method is proposed to extract the characteristic information of UHF PD, and the UHF PD signals from four typical discharge models obtained in the laboratory are analyzed. In order to overcome the defect of spectrum aliasing and energy leakage among real wavelet packet decomposition subbands, HWPT is used to solve the problem of multi-scale decomposition. The difference of energy and complexity of UHF PD signal at different scales is used. The multi-scale energy and multi-scale sample entropy parameters are extracted as characteristic variables for pattern recognition, and the time-frequency domain information of UHF PD signal is described more accurately. Finally, the results of SVM classification and recognition show that this method can achieve better recognition effect than real wavelet packet, and multi-scale energy and multi-scale sample entropy feature parameters can effectively identify four kinds of insulation defects.
【作者单位】：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;
【基金】：国家重点基础研究发展计划(973计划)(2009CB724506) 国家自然科学基金(5177181)资助项目
【分类号】：TM595

【相似文献】