结合H-S变换和BSVM的高压输电线路故障识别

High-voltage Transmission Line Fault Identification Based on the Hyperbolic S-transform and Multiple BSVM

针对高压输电线路故障识别元件易受系统工况和一些不确定因素影响的问题,提出一种新的故障分类模型。此模型主要由3个二分类支持向量机分类器、1个零序分量判别器和1个逻辑判断器构成。对三相电流进行H-S变换得到模时频矩阵值,用于对支持向量机分类器的训练;对零序电流进行小波变换计算低频能量,用于零序分量判别器识别;最后逻辑判断器根据二者的输出遍历逻辑表做出判断。通过仿真实验,此模型在各种干扰和工况变化的情况下都能保持良好的性能。

The element for high-voltage transmission line fault identification is easily affected by the working condi- tions of power system and some uncertainties. An improved fault classification model is presented. This model is mainly composed of three binary support vector machine （BSVM）, a zero-sequence current checker （ZSCC）and a logical judgment component （LJC）. For three-phase current, hyperbolic S-transform is used to get modular values of the time- frequency matrix which are adopted to train BSVM. Wavelet transform on the zero-sequence current is used to calculate the low-frequency energy, which is the input to the ZSCC. On the basis of their output, the LJC finally traverses a logic table to get the result. Through the simulation, this model can maintain good performance in the cases of various distur- bances and working conditions.