The Simulation Of Inrush Current Of Transformer And Research Of Recognition Algorithm

As a key device, power transformer is the vital important electric installation in electric system. At present, the action of transformer protection is in a low accuracy, it affects the security and stability of the grid. Longitudinal differential protection is one of the main transformer protection, the correct identification of transformer magnetizing inrush from internal short-circuit fault is a key issue of transformer protection. This paper focuses on the study of methods to identify magnetizing inrush current.The paper introduces the development and actuality of transformer simulation and protection, analyzes the mechanism of magnetizing inrush current for transformer and its effect on differential protection.To study the digital method of simulation for transformer, the models of inrush current and internal fault for transformer have been established by ATP. The transformer fault is simulated under different working conditions.Based on analyzing the waveform difference of inrush and fault current, a Non-sinusoidal fractal estimation method is proposed to identify inrush current. It employs grille fractal technique to compute Non-sinusoidal.It is stable during the current of different amplitude, and has little relation with the parameter of the transformer, so the method is able to offer high reliability and is easy to implement.Under transformer internal and external faults, the curvature of differential current is calculated and analyzed. Based on the curvature characteristics of waveform, a novel approach to identify inrush current and internal fault current is presented. Considering curvature calculation is sensitive to the effect of noise, to ensure the reliability of the curvature algorithm, morphological filter with self-adaptive method is used to filtrate noises and interference signals.By the simulation of ATP and MATLAB, both methods are feasibility and effective to identification inrush for transformer.