Modular multilevel converter (MMC) has been widely used in wind power system. The design of insulated gate bipolar transistor (IGBT) directly affects the performance of the system. Because of the lack of theoretical guidance in the design and selection of IGBT modules, there is a phenomenon of cost wasting which is caused by a large margin selection. Based on this, a customized design method for IGBT based on dual objective optimization is proposed. Firstly, the cost and loss of the IGBT are taken as the goals of the optimization. Combined with the annual operating conditions of wind power generation system, IGBT loss is evaluated according to the probability distribution of wind speed. Then, the Pareto front is obtained by dual-objective particle swarm optimization algorithm. Finally, customized design for IGBT module and discrete components are used to design the converter. The results show that the proposed design method reduces the cost of the device and takes into account the loss and reliability of the converter at the same time.