To analyze the inrush current in a superconducting transformer, the machine parameters for the transformer were estimated from the measured current using a search algorithm. To address the large rising edge error in estimations performed using a genetic algorithm (GA), a differential evolution (DE) was used in this study. As a result, the estimated time was reduced to about 1/10 that obtained with GA, and the evaluation value indicating the difference between the measured value and the estimated value was reduced to about 1/2. Thus, it was possible to estimate with higher accuracy by using DE.