采用广义S变换的电能质量扰动免疫分类算法

Power Quality Disturbances Classification Using Generalized S-transform and Artificial Immune

针对电能质量扰动分类问题,提出了一种基于广义S变换和人工免疫算法相结合的电能质量扰动分类新方法。给出了广义S变换的定义,对比了信号经广义S变换和S变换后的变换结果,推导了一维广义S变换实现过程,给出了调节因子λ随信号中频率组成不同而进行自适应取值的方法。分类过程中,首先使用广义S变换提取出各种扰动信号的特征量形成抗原,然后对不同扰动样本按照提出的免疫分类算法步骤进行训练形成可以分类的抗体,计算抗体和待分类的抗原之间的欧氏距离并用最近邻法则输出最终分类结果。仿真实验结果表明广义S变换比S变换精度高,采用的分类方法能够实现电能质量扰动的自动分类,且对噪声不敏感,分类正确率很高。

A new approach combining generalized S-transform （GST） and artificial immune classifier was proposed to classify the power quality disturbances. Definition of GST was described, the signal analysis results of GST and the S-transform were compared, the computing formula was derived, and the implementation process of GST was de- scribed. Moreover, an adaptive method to adjust the parameter 2 was designed to satisfy the different resolution needs, which adjusts λ according to the frequency components of the signal. During the classification processing, firstly, some main power quality disturbances were analyzed, their GST curves were drawn and five disturbances feature components were extracted as antigens of the artificial immune system, then different disturbance samples were trained to get the antibodies by using the artificial immune classification algorithm, finally, space distance be- tween the antibodies and antigens were computed and the classification was performed in k-nearest neighbor ap- proach. The simulation was done in Matlab environment, 150 samples of every disturbance were selected as the training group and the other 150 samples were selected as testing group. The simulation results indicate that the GST has higher precision than the S-transform, the proposed method can classify the power quality disturbances ef- fectively, and it has high classification correct ratio and is not sensitive to the noise.