基于磁场和支持向量机的空心电抗器匝间短路诊断研究

Research on Interturn Short Circuit Diagnosis of Drytype Aircore Reactor Based on Magnetic Field and Support Vector Machine

针对干式空心电抗器的匝间短路故障检测问题,提出了一种利用电抗器磁场幅值与支持向量机的匝间短路故障诊断方法,并仿真验证其有效性。首先建立了干式空心电抗器的场路耦合模型,研究分析了电抗器在不同故障条件下底部磁场幅值和角度分布的变化情况,并以磁场幅值与角度的变化信息为特征数据,提出了基于SSA-SVM的故障分类诊断模型,结合实际应用中的限制条件,磁场分布选取均匀的测量位点处的磁场数据所组成的向量进行表示,并引入欧式距离这一数学指标描述其变化。仿真结果表明,磁场信息能够有效适用于SVM模型进行匝间短路故障的分类诊断,且通过麻雀搜索算法(sparrow search algorithm,SSA)对其惩罚因子和核参数进行寻优后,得到的SSA-SVM模型获得了更好的诊断准确率。

In view of interturn short circuit fault detection of dry air core reactor,a kind of interturn short circuit faults detection method using the magnetic field amplitude and the support vector machine is proposed and its effectiveness is verified by simulation.Firstly,the field circuit coupling model for a dry type aircore reactor is set up,the variation in the amplitude and angle distribution of the bottom magnetic field of the reactor under different fault conditions is studied and analyzed and the the variation information of magnetic field amplitude and angle are taken as the characteristic data,the fault classification and diagnosis model based on SSASVM is proposed.To consider the limitations in practical applications,the magnetic field distribution is represented by a vector which is composed of magnetic field data at uniform measurement points,and the Euclidean distance is introduced as a mathematical indicator to describe its variation.The simulation results show that the magnetic field information can be effectively applied to the SVM module for classification and diagnosis of interturn short circuit faults,and the obtained SSASVM model achieves better diagnostic accuracy after optimizing its penalty factor and kernel parameters through SSA.