基于FST和LibSVM的电能质量扰动信号分类  

Power Quality Disturbance Signal Classification Based on FST and LibSVM

作　　者：孙玉杰 张占强[1] 孟克其劳 吕晓圆 SUN Yu-jie;ZHANG Zhan-qiang;MENGKE Qi-lao;LV Xiao-yuan(School of Information Engineering,Inner Mongolia University of Technology,Hohhot Inner Mongolia 010080,China;College of Energy and Power Engineering,Inner Mongolia University of Technology,Hohhot Inner Mongolia 010080,China)

机构地区：[1]内蒙古工业大学信息工程学院,内蒙古呼和浩特010080 [2]内蒙古工业大学能源与动力工程学院,内蒙古呼和浩特010080

出　　处：《计算机仿真》2022年第11期146-152,共7页Computer Simulation

基　　金：国家自然科学基金(51467016);内蒙古自治区自然科学基金面上项目(2015MS0618);内蒙古自治区高等学校科学研究项目(NJZY080)。

摘　　要：针对电能质量扰动信号种类多,特征提取精度不高,运算时间长,难以被正确分类等问题,提出一种快速S变换和LibSVM的电能质量扰动信号分类算法。在FST基础上引入调节因子提高信号时频分辨率,利用快速S变换提取扰动信号特征向量,将此向量分为训练样本和测试样本,归一化处理后输入到LibSVM进行分类,与S变换相比,具有节省运算时间,分类准确率高等优点。仿真结果表明,在样本较少的情况下加入不同信噪比的噪声,分类所用时间短、正确率高、抗干扰能力强,适合于电能质量扰动信号分类。Aiming at the problems of many kinds of power quality disturbance signals,low accuracy of feature extraction,long operation time and difficult to be classified correctly,a fast S-transform and LibSVM algorithm for power quality disturbance signal classification is proposed.On the basis of FST,the adjustment factor was introduced to improve the time-frequency resolution of the signal.The feature vector of the disturbance signal was extracted by using fast S transform.The vectors were divided into training samples and test samples.After normalization,it was input to LibSVM for classification.Compared with S transform,it has the advantages of saving operation time and high classification accuracy.The simulation results show that the time of classification is short,the accuracy is high,and the anti-interference ability is strong when the samples are small,which is suitable for power quality disturbance signal classification.

关 键 词：电能质量扰动信号 特征提取 分类 

分 类 号：TP391.5[自动化与计算机技术—计算机应用技术]