摘要
在研究二叉树多分类的基础上,结合输电线路故障的特点,考虑不同故障的优先级后,设计了基于SVM的改进二叉树输电线路故障分类器的模型,通过实验选择了最小二乘支持向量机LS-SVM算法和线性函数转换表达式的归一化算法,并用小样本模拟短路数据训练了分类器。测试结果表明,在各种输电线路故障情况下,设计的分类器都具有很高的分类正确率,尤其是对两相接地和不接地短路分类的效果显著,另外,该分类器的数据预处理过程简单,分类步骤少,可以实现输电线路故障的快速分类。
Based on SVM an improved binary tree methods multi-classifier algorithm model for transmission line faults is designed in which the characteristics and the priorities of different faults of transmission line are considered, and the least squares algorithm for SVM and the data normalization algorithm named linearity function transform expression for data pretreatment are selected according to the results of experiments. Then the multi-classifier is trained by several simulated short-circuit samples. The test results show that the correct rate of the classifier is very high for all kinds of transmission line faults, especially significant high for two-phase short-circuit and two-phase to ground fault. In addition, the classification process for transmission line faults is fast because of the simple data pretreatment and the reduced classification steps.
出处
《电力系统保护与控制》
EI
CSCD
北大核心
2010年第5期39-44,共6页
Power System Protection and Control
基金
湖北省教育厅自然科学研究项目计划(D20091304)