A hybrid optimization algorithm(HOA) is proposed based on genetic algorithm(GA) and simplex method. The searching abilities of optimization algorithms with different coding methods are compared. The results show that the searching abilities of HOA is remarkable performance against that of GA; The real-coded HOA can maintain the population diversity and is competitive with the binary-coded HOA when solving problems of many local optimal solution. In order to reducing the aerodynamic drag of high-speed trains, the cross-sectional area distribution of a high-speed train nose is optimized with the real-coded HOA combined with Hicks-Henne function parametric method and Kriging surrogate model, and the best cross-sectional area distribution in the design space is found. The aerodynamic drag of the original shape is reduced by 9.41%, the viscous drag is reduced by 38.02%, the inviscid drag change little, the aerodynamic drag of the nose and the trailing car is reduced by 12.55% and 13.98%, respectively.