The J-A hysteresis theory is widely used in the modeling of the magnetic properties of electromagnetic current transformers. However,under harmonic conditions,the model has significant errors. To solve this problem,the classical J-A model parameters are modified for each harmonic,and genetic-annealing algorithm is used for parameter identification. Then,the magnetic characteristics of each harmonic are linearly added to achieve the magnetic characteristics modeling of the current transformer under non sinusoidal excitation. The premise of adopting this correction method is that the nonlinear error of the measuring current transformer meets the specified level 0.2S,so the simulation analysis and error correction of harmonic hysteresis loops are approximately applicable to the linear superposition principle. Taking 0.2S level current transformers made of nanocrystalline materials as an example,the simulation analysis and experimental verification of angle errors and ratio errors show that the J-A model after parameter correction improves measurement accuracy under harmonic conditions,and the approximate linear superposition method of multiple harmonic magnetic characteristics has a negligible impact on measurement accuracy. This confirms that the J-A hysteresis model correction considering harmonics is reasonable and effective.