基于SVM的电能质量扰动分类

Research of Power Quality Disturbance Classification Based On SVM

电能质量扰动信号数据庞大,数据提取较难。本文对常见的电能质量扰动及其组合的复合扰动进行离散小波分解,提取PQ(Power Quality)扰动信号能量差作为特征向量,以此降低扰动分类的数据量。利用MATLAB软件产生PQ扰动训练和测试样本,在扰动样本中加入SNR=25d B的高斯白噪声,利用SVM对扰动样本进行分类,提出两步网格搜索法对SVM的参数进行优化。仿真实验结果表明,此分类方法具有较高识别率,证明该算法的准确性和鲁棒性。

There is much data about the power quality disturbance signals in the power system, so it is difficult to ex-tract the data.This paper adopted a method of discrete wavelet transformation to discompose the common power quality disturbances and the compound disturbances of their combination.This paper extracted the energy subtraction of the power quality disturbance signals as the signal feature vector so as to reduce the data of disturbance classification.The software of MATLAB was used to generate the disturbance signal samples and the Gauss white noise with the signal-noise ratio of 25 dB was added in the samples.Then the SVM was used to classify the disturbance samples and a two-step grid search method was proposed to optimize the SVM parameters.The simulation results showed that the proposed method had a very high identification rate and they also verified the correctness and robustness of this method.