基于支持向量机的复杂环境条件下绝缘子闪络电压的预测

Insulator Flashover Voltage Forecasting Under Complex Circumstance Based on Support Vector Machine

在大型人工气候实验室对XZP-160绝缘子试验数据的基础上,提出了一种基于支持向量机的绝缘子闪络电压预测方法。支持向量机是以统计学习理论为基础的,采用结构风险最小化原则代替传统经验风险最小化原则的新型统计学习方法。该文以气压、覆冰、污秽程度等环境条件作为输入,绝缘子的闪络电压作为输出,对环境条件和闪络电压的关系进行训练,建立绝缘子闪络电压的预测模型。结果表明预测的闪络电压与实测结果基本一致。该方法为复杂环境条件下外绝缘的选择提供了一种新的途径。

According to the test result on XZP-160 insulator in a large artificial climate chamber, a flashover voltage forecasting method based on support vector machine （SVM） is put forward. It is a new statistical study method in which the traditional empirical risk minimization principle is replaced by structural risk minimization principle. Using environmental conditions （atmosphere pressure, ice weight and pollution degree） as inputs, insulator flashover voltage as outputs, the relation between environmental conditions and flashover voltage is trained and the flashover voltage forecasting model is built. The forecasting result is in concordance with test result. The method provides a new way to select external insulation under complex circumstance conditions.