Research on Identification of Magnetizing Inrush Current Based on PSO-SVM

Abstract

When the transformer is put into operation, the reliable operation of protection is affected by residual magnetism and magnetic bias. The traditional second harmonic differential protection is difficult to distinguish the inrush current and the fault current under the special angle of the transformer closing angle and the certain value of the remanence, which results in the transformer protection not moving or refusing to operate. This paper proposes an optimized support vector machine (SVM) magnetizing inrush current recognition model based on the particle swarm algorithm (PSO), and combines the current waveform itself to perform feature extraction to realize the identification of transformer magnetizing inrush current and internal faults. The influence of closing angle and remanence on inrush current is studied and the effective feature is extracted to identify the inrush current intelligently. The experimental results show that the extracted feature and identification method are effective.