应用原子分解的电能质量扰动信号分类方法  

Classification of Power Quality Disturbance Signals Using Atomic Decomposition Method

作　　者：王宁[1] 李林川[1] 贾清泉[2] 董海艳[2] 

机构地区：[1]电力系统仿真控制教育部重点实验室(天津大学),天津市南开区300072 [2]燕山大学电气工程学院,河北省秦皇岛市066004

出　　处：《中国电机工程学报》2011年第4期51-58,共8页Proceedings of the CSEE

基　　金：国家自然科学基金项目(50877069)~~

摘　　要：根据信号原子分解的思想,针对电能质量扰动信号的特点构建4种相关原子库,即基波原子库、类基波原子库、脉冲原子库和振荡原子库。通过改变搜索参数的方式,实现对匹配追踪算法的优化。首先通过匹配追踪算法提取基波分量,对去除基波分量的残余信号提取类基波扰动,并修正基波分量和类基波扰动,然后提取振荡扰动,最后提取脉冲扰动,得到基波和各种扰动的参数。根据扰动分量能量与残余信号能量之间的关系构造分类方法。仿真和实测电能质量扰动录波数据表明,该分类方法具有分类准确率高、对噪声不敏感等优点。In this paper, using the atomic decomposition method, four coherent dictionaries were designed based on analyzing characteristics of power quality disturbance signals. The matching pursuits （MP） algorithm was optimized by changing searching parameters. The fundamental frequency component, similar fundamental frequency disturbance, oscillatory disturbance and pulse disturbance were extracted from the original signal in sequence using the MP algorithm. During this process, the fundamental frequency component and similar fundamental frequency disturbance were revised before extracting oscillatory disturbance and pulse disturbance. The parameters of the fundamental frequency component and disturbances were gained. According to the relationship of signal energy between the disturbance components and the residual of wiped off fundamental component, the disturbance types of power quality could be classified. Simulation and real data results show that the classification method has high accuracy and signal-to-noise ratios.

关 键 词：电能质量 扰动分类 原子分解 匹配追踪 相关原子库 

分 类 号：TM72[电气工程—电力系统及自动化]