In order to realize fast identification of short-circuit fault current and accurate prediction of zero-crossing point of variable impedance transformer, the short-circuit current identification method based on curvature algorithm of current waveform and the zero-crossing prediction algorithm of short-circuit current based on least square method are studied. The influence of initial phase angle and fundamental frequency angle of different faults on zero prediction under ideal condition, different signal-to-noise ratio and different harmonic conditions is analyzed,and a short-circuit fault model is built with MATLAB and simulated. The results show that the short-circuit current identification method based on curvature algorithm of current waveform and the zero-crossing point prediction algorithm based on least square method can identify fault current quickly, and greatly reduce the prediction error of zero-crossing point time of short-circuit current.