Abstract: There exist the defects of multi design variables and long optimization cycle in the traditional optimum
design of compressor blades． In order to solve these problems，this paper proposes a free-form mesh deformation parameterization method of fluid grids by using a non-uniform rational B-spline basis function，and constructs an design system of aerodynamic optimization of rotor blades by combining the advanced design of Latin hypercube sampling experiments，the Kriging response surface model and the NSGA-Ⅱmulti-objective genetic algorithm． Calculation results show that (1) the optimized blade has a total pressure ratio improvement by 0．33% and an isentropic efficiency improvement by 0．83% at a choke mass flow of 98%; (2) the optimized blade is a forward-leaned blade，which helps reduce the shock loss and improve the performance of the rotor; and (3) in comparison with the traditional optimization design method，the proposed optimization system reduces design variables and shortens optimized cycles．

Key words: compressor, optimum design, mesh deformation, response surface, multi-objective genetic algorithm