中文摘要：

提出一种基于多分辨奇异谱熵和支持向量机的特高压直流输电线路区内外故障识别方法,可准确将本侧区外故障、区内故障以及对侧区外故障区分开。进行小波多尺度分解,求得各层的奇异谱熵,将每层的奇异谱熵组成一个特征向量,特征向量分成训练集和测试集,将训练集进行训练得到支持向量机（support vector machines,SVM）分类器的参数,用测试集进行测试,预测结果就是对不同位置故障的分类。大量仿真验证表明：基于多分辨奇异谱熵和支持向量机的特高压直流输电线路区内外故障识别方法能可靠识别本侧区外故障、区内故障和对侧区外故障。

英文摘要：

Based on multi-resolution singular spectrum entropy and supporting vector machine （SVM）, a transient protection principle method for UHVDC transmission is proposed, by which the faults located outside or inside the protection zone of the local side can be accurately distinguished from the fault located outside the protection zone of the opposite side. The singular spectrum entropy of different layers is obtained by multi-scale wavelet transform decomposition, and the characteristic vector of each layer is composed of the singular spectrum entropy of each layer, which is divided into training set and test set. The training set is trained to achieve the parameters of SVM classifier, which are tested by the test set, and the prediction results are the classification of the faults located at different positions. A lot of simulation results show that the multi-resolution singular spectrum entropy and SVM based method can reliably recognize the faults located outside or inside the protection zone of the local side and the fault located outside the protection zone at opposite side as well.