The 1 000 kV dry-type air-core shunt reactor is a key device in ultra-high voltage AC transmission systems. However

research on its grading ring configuration remains relatively limited. To reduce the maximum electric field intensity of the reactor

this study proposes an optimization method based on the combination of Bayesian optimization (BO) and the Kriging model. By optimizing the hyperparameter of the Kriging model via the BO algorithm

a high-precision BO-Kriging surrogate model is constructed. The results demonstrate that

compared with polynomial response surface

radial basis function response surface

support vector regression

and Kriging model

the Kriging model exhibits superior predictive capability and fitting accuracy; further comparison of hyperparameter optimization methods

namely

grid search

random searching

genetic algorithm

and BO algorithm

reveals that the BO-Kriging model significantly reduces root-mean-square error

mean absolute error

mean absolute percentage error

and the relative error between predicted and simulated maximum field intensity in 10-fold cross-validation. After optimization

the maximum electric field intensity of the reactor is reduced by 75.903%

validating the effectiveness of the BO-Kriging model. The proposed method can be adopted to not only ensure the fitting accuracy of the surrogate model

but also drastically reduce the maximum electric field intensity

meeting engineering requirements. This study provides a high-precision optimization framework and theoretical foundation for the grading ring design of the 1 000 kV dry-type air-core shunt reactor.