As high-frequency transformers usually works under complex excitation conditions, when using the traditional AP method to design it, it is often impossible to balance the relationship between multiple goals at the same time. The calculation method for core loss, winding loss and the product of area method are used as the objective functions, the efficiency of transformer is taken as the constraint condition, loss of the transform reduction and improve efficiency as the main objective in this paper. An advanced MOTO algorithm is adopted to carry out the multi-objective optimization design of the high frequency transformer. By adopting adaptive mutation and crossover probability to adjust the diversity of the population and adding the Pareto dominance relationship to improve the superiority of the population, the algorithm’s global search capability is improved. And the improved MOTO algorithm can provide support for solving multi-objective problems with constraints. Moreover, the entropy weight method (EWM) is introduced to support the decision-making of the algorithm optimization results, which is convenient for decision-makers to choose the best compromise solution and obtain a high-frequency transformer optimization design scheme. Finally, experiments verify the rationality and effectiveness of this scheme.