Abstract: In order to reduce the circulating current loss, in view of the characteristics of long optimization cycle and low optimization efficiency in the conductor transposition optimization design of power transformer, a parallel Kriging model global optimization algorithm based on two points addition criteria is proposed. On the one hand, to improve the prediction accuracy of the optimum, a local infilling criterion with adaptive scaling is proposed. On the other hand, based on the characteristics of predicting mean square error combined with the concept of confidence interval, an adaptive global infilling criterion is proposed. The parallel infilling criterion is used in parallel to improve the predicting accuracy of Kriging model while improving the optimization efficiency. The effectiveness of algorithm is validated by some analytic functions and TEAM 22. Finally, the proposed method is applied to an optimal transposition design of a power transformer to minimize the circulating current loss.