SCG算法在GIS设备局部放电模式识别中的应用研究  

Research on the Application of SCG Algorithm in Partial Discharge Pattern Recognition of GIS Equipment

作　　者：邢雅 侯峰 吴培涛 冯洋 王宏 XING Ya;HOU Feng;WU Peitao;FENG Yang;WANG Hong(Training Center of State Grid Ningxia Electric Power Co.,Ltd.,Yinchuan 750000,China)

机构地区：[1]国网宁夏电力有限公司培训中心,宁夏银川750000

出　　处：《微型电脑应用》2024年第2期97-100,共4页Microcomputer Applications

基　　金：国网宁夏电力有限公司培训中心项目(SGNXPX00JNJS2310055)。

摘　　要：为了更好识别气体绝缘组合电器设备复杂的局部放电模式,利用特高频检测法和比例共轭梯度算法检测局部放电信号,并搭建气体绝缘组合电器设备模型。结果表明,当时间为13.5 ns时,检测到2号传感器的放电信号,并且波形频率更高说明离放电电源更近。在训练集,悬浮与尖端放电的识别准确率为80.6%,复杂局部放电类型平均识别准确率为86.2%。在验证集,悬浮与尖端放电的识别准确率为83.3%,平均准确率为93.2%。设置30的隐含层数,此时数据分类的准确率为87.3%。这证明局部放电类型识别准确率较高,对检测电力系统的安全性和可靠性提供保障。In order to better identify the complex partial discharge patterns of gas insulated composite electrical equipment,the ultra high frequency detection method and proportional conjugate gradient algorithm are used to detect the partial discharge signals,and a model of gas insulated composite electrical equipment is built.The results indicate that when the time is 13.5 ns,the discharge signal of sensor 2 is detected,and a higher waveform frequency indicates a closer proximity to the discharge power source.In the training set,the recognition accuracy of suspension and tip discharge is 80.6%,and the average recognition accuracy of complex partial discharge types is 86.2%.In the validation set,the recognition accuracy of suspension and tip discharge is 83.3%,with an average accuracy of 93.2%.When we set the number of hidden layers to 30,the accuracy of data classification is 87.3%.It has been proven that the recognition accuracy of partial discharge types is high,which provides a guarantee for the safety and reliability of detecting power systems.

关 键 词：GIS SCG算法 局部放电 特高频检测法 特征提取 

分 类 号：TP389.1[自动化与计算机技术—计算机系统结构]