基于特征组合的SVM电能质量扰动信号分类

针对电能质量扰动信号(PQD)特性,讨论了PQD信号分类时的特征选择,提出了基于多特征组合进行PQD分类的新思路。提取PQ测度、小波能量和均方根值(RMS),将这三种单项特征组合形成组合特征,使用支持向量机进行分类,比较了单项特征与组合特征分类的各项参数,确定最佳特征组合,实验中采用Matlab数学模型仿真数据、Fluke 6100A标准源数据、电能质量监测网现场采集数据分别进行实验,验证了该思路的可行性和正确性。

Power quality disturbance signal classification using support vector machine based on feature combination

In view of power quality disturbance(PQD)characteristic,the author discussed the feature selection of PQD classification and proposed a new method of PQD classification which adopts features combination.In this paper,first the power quality measurement,wavelet energy and RMS were calculated,and the combination feature was combined.Then SVM was used to class.At last,the correlative parameter of single feature and combination feature was compared in order to select the best combination feature.In the experiment,the experiment data came from three aspects which are emulation data by math model,data generated by Fluke 6100 and data from the local collected by power quality monitor net.The results indicate that the new method above is right and feasible.