一种基于二分类的电压暂降源定位方法  

A two-class classification based method for voltage sag source location

作　　者：吕干云[1] 孙维蒙[1] 

机构地区：[1]浙江师范大学数理与信息工程学院,浙江金华321004

出　　处：《电力系统保护与控制》2012年第13期84-87,共4页Power System Protection and Control

基　　金：国家自然科学基金项目(51107120);浙江省自然科学基金项目(Y1090182);浙江师范大学计算机软件与理论省级重中之重学科开放基金;浙江省教育厅科研项目(Y201120550)~~

摘　　要：电压暂降源定位实质是根据暂降源位置敏感的物理量,来确定暂降源位于监测点的上游或下游位置,模式识别上属于二分类问题。提出了一种基于智能分类的电压暂降源定位方法。分析回顾了现有的几种典型暂降源定位方法及其判据,通过判据研究提取源位置敏感的多个定位特征量,并采用支持向量机智能算法对已有故障诊断结果进行学习,从而在该多维特征空间构建判别上下游的最优分类面,利用此最优分类面(即为二分类支持向量机)实现基于二分类的暂降源定位。测试数据表明,该方法能够有效实现电压暂降源定位,定位准确率高且所需计算时间短。Voltage sag source location is to determine that the sag source is located upstream or downstream relative to the monitoring points based on the physical quantity sensitive to voltage sag location. From the view of pattern recognition, it is a problem of two-class classification. A new method is developed to locate the sag source based on thought of smart classification. Several typical existing sag source location methods and their criterions are reviewed first. Many features sensitive to source location are extracted according to the above criteria. Using support vector machine （SVM） intelligent algorithm, the obtained fault diagnosis results are trained and an optimal hyper-plane is constructed in the space of multi-features to distinguish upstream from downstream. Finally, the optimal hyper-plane, that is the trained SVM, is used for two-class classification based voltage sag source location. The testing results show that the proposed method can locate the origin of voltage sag effectively, and shows a high accuracy of location and short calculating time.

关 键 词：电压暂降源 定位 二分类 支持向量机 最优分类面 

分 类 号：TM714.2[电气工程—电力系统及自动化]