To further improve the computational speed and accuracy of electromagnetic force in transformer windings

this paper proposes a winding electromagnetic force calculation method based on the second-order magnetic field tensor

and the electromagnetic force distribution in a 500 kV converter transformer during a short-circuit fault is adopted as a case study. First

the primary distribution pattern of the magnetic field in the winding is determined by using the first-order tensor B of the magnetic flux density in the Cartesian coordinate system

and its correlation with the electromagnetic force F is verified. Next

the scalar distribution of B and the form of the second-order tensor are further studied. Correlation analysis reveals that the variation in the magnitude and direction of B is in complete agreement with the distribution characteristics of the electromagnetic force. Subsequently

the relationship between the second-order tensor and the electromagnetic force is fitted by using the gradient boosting regression tree model

and the optimal hyperparameters are determined through Bayesian optimization. Finally

a 35 kV scaled-down transformer test platform is constructed

and the measured electromagnetic force is compared in detail with the fitted values based on short-circuit test results. The relative error does not exceed 1.8%

validating the effectiveness and universality of the proposed method. This approach simplifies the calculation process and enhances computational efficiency while maintaining accuracy

providing significant reference value for transformer design and analysis with broad engineering applicability.