An efficient optimization design method for centrifugal compressor blades

To improve the time-consuming problem of centrifugal impeller optimization design, an efficient optimization design method is proposed. A bionic evolutionary algorithm based on a multi-surrogate model is established, model management strategies and geometric properties of the sigmoid function are investigated, and the penalty mechanism of infeasible solutions is revealed, which finally increases the speed of iterations and improves the optimum solution of constraint optimization problem for centrifugal compressor blades. Combining multi-surrogate model evolutionary algorithm, surface parameterization method, and CFD for aerodynamic optimization of blade shape, the optimization results shows that the isentropic efficiency increased by 1.9%, the mass flow rate increased by 4.61%, the total pressure ratio increased by 0.81%, and the computational time was reduced by 54.9%. The sigmoid-based multi-surrogate model evolution algorithm improves the isentropic efficiency by 1.19% and the total pressure ratio by 0.63% compared with the unconstrained multi-surrogate model evolution algorithm, while the sigmoid-based multi-surrogate model evolution algorithm improves the isentropic efficiency by 1.16% and the total pressure ratio by 0.55% compared with the feasible point dominance constrained multi-surrogate model evolution algorithm, verifying the efficiency of the sigmoid-based multi-surrogate model evolution algorithm.