Abstract: In order to realize the rapid calculation of the dynamic magnetic field of transformer during the short circuit, an adaptive nonlinear degree selection of extended dynamic mode decomposition for control (ANDS-EDMDC) method was proposed. Compared with the extended dynamic modal decomposition method, ANDS-EDMDC introduces control terms and adaptive nonlinear degree selection strategies, which further improves the applicability and accuracy of dynamic magnetic field calculation. Firstly, in order to verify the effectiveness of ANDS-EDMDC, the full-order simulation and computational analysis of the dynamic magnetic field were carried out with the normal operation state of TEAM21 as the research object. Based on the magnetic field simulation results of COMSOL, the calculation error of the improved model is within 3.698%, which proves the correctness of the proposed method. Secondly, the results of the improved model were compared with the actual values, and the results showed that the improved model had a smaller error compared to the actual values. Finally, ANDS-EDMDC was applied to analyze the transient magnetic field during the three-phase short-circuit process of a transformer, and compared with the dynamic mode decomposition method and the extended dynamic mode decomposition method. The results show that the maximum average relative error of EDMD is about 14.7 times that of ANDS-EDMDC, and the algorithm can improve the efficiency of simulation calculation by about 2 263 times.