Performance Optimization Design of Diagonal Flow Fan Based on Ensemble of Surrogates Model

Abstract

Due to the advantages of a high total pressure coefficient, large flow coefficient, and high efficiency, the diagonal flow fan is widely used in people’s livelihood and industrial fields. However, the design of the diagonal flow fan is mostly empirical, multi-solution, and comprehensive. The traditional optimization design process often consumes huge computing resources. In this paper, the diagonal flow fan blade is parameterized, the design variables are determined, and the accuracy of the parameterization method is verified. The maximum fitting error is controlled at approximately 0.1%. Based on the parametric design of blades, this paper organically integrates the traditional Kriging model and RBF model, and introduces the Ensemble of surrogates model (ES) to verify that the ES model has higher prediction accuracy in the prediction of fan flow and total pressure efficiency than the traditional prediction model. Subsequently, the Pareto optimal solution set of the approximate model within the global design scope is searched by NSGA-II. The numerical simulation and experimental verification show that the actual flow of the fan increases by 10% and the efficiency of the full pressure increases by 3.2% under the design condition of the optimized blade. The optimized model can significantly improve its air performance.