为了提高过电压识别的准确性，以及加快过电压的识别速度，提出一种自适应遗传算法优化支持向量机的过电压识别方法。首先针对单一特征信息难以获得过电压高识别率的问题，采用时域波形、波头、时频谱的组合特征作为过电压识别特征，然后采用过电压的训练样本对支持向量机进行学习，建立过电压识别的分类器，并引入自适应遗传算法对支持向量机参数进行优化，最后采用具体过电压识别实例进行性能仿真分析。结果表明，该方法的过电压平均识别率达到95%以上，远远超过了实际应用的85%要求，且识别结果要优于其他过电压识别方法。%In order to improve the overvoltage recognition accuracy and quicken the recognition speed,an overvoltage iden⁃tification method with genetic algorithm optimizing support vector machine is proposed. Since the single feature information is dif⁃ficult to obtain the high overvoltage recognition rate,the combined features of time domain waveform,wave head and time⁃fre⁃quency spectrum are taken as the recognition features of overvoltage respectively,and then the training samples of overvoltage are used to study the support vector machine. The classifier of overvoltage identification is established,and the adaptive genetic algorithm is introduced to optimize the parameters of support vector machine. The performance of an overvoltage recognition in⁃stance was performed for simulation analysis. The results show that the average overvoltage recognition rate of the proposed method can reach up to 95%,far exceeds the practical application requirements of 85%,and the recognition result is superior to other overvoltage identification methods.
展开▼
