基于PRPD图谱的气体绝缘开关柜多缺陷局部放电模式识别  

Partial discharge pattern recognition of multiple defects in gas insulated switchgear based on PRPD spectrum

作　　者：冯旗 邵振华 余祉宏 FENG Qi;SHAO Zhenhua;YU Zhihong(College of Electrical and Electronic Engineering Wenzhou University,Wenzhou 325035,China;College of Computer and Control Engineering Minjiang University,Fuzhou 350108,China)

机构地区：[1]温州大学电气与电子工程学院,浙江温州325035 [2]闽江学院计算机与控制工程学院,福建福州350108

出　　处：《电气应用》2023年第6期48-54,共7页Electrotechnical Application

基　　金：福建省科技厅引导性项目资助(编号:2021H00055和2021H01010009)。

摘　　要：针对目前不断提高的电压等级,局部放电现象不再是单独一种,可能存在多缺陷同时放电的情况,局部放电类型的模式识别,对监测设备的运行状态具有重要意义。然而多缺陷所产生的局部放电信号存在混叠现象,给识别带来困难。为了能够识别多缺陷局部放电,从PRPD图谱入手,提取图谱的矩特征和灰度共生矩阵特征,将其组合后的7维特征采用支持向量机算法进行分类。该方法可有效解决多缺陷放电信号混叠导致的传统信号特征无法提取的问题。搭建实验模型,共采集了四种单一典型缺陷和三种混合缺陷放电的PRPD图谱,比较了矩特征、灰度共生矩阵特征和组合特征的识别率,结果表明利用组合特征不仅在识别速度上有较大的提升,而且识别率也能够得到提高,可以有效地识别出单一缺陷和混合缺陷,验证了该混合特征的有效性。With the increasing voltage level,partial discharge is no longer a single phenomenon,and there may be simultaneous discharge of multiple defects.The pattern recognition of partial discharge types is of great significance for monitoring the operation status of equipment.However,the partial discharge signals generated by multiple defects are overlapped,which makes it difficult to identify.In order to recognize multi defect partial discharge,this paper starts with PRPD atlas,extracts the moment features and gray level co-occurrence matrix features of the atlas,and classifies the combined 7-dimensional features with support vector machine algorithm.This method can effectively solve the problem that the traditional signal features cannot be extracted due to the overlapping of multi defect discharge signals.The experimental model was built.The PRPD maps of four single typical defects and three mixed defects were collected,and the recognition rates of moment features,gray level co-occurrence matrix features and combined features were compared.The results showed that using combined features not only improved the recognition speed,but also the recognition rate could be improved,which could effectively identify single defects and mixed defects.The effectiveness of the mixed features was verified.

关 键 词：多缺陷局部放电 PRPD图谱 矩特征 灰度共生矩阵 支持向量机 

分 类 号：TM591[电气工程—电器] TM855