Applied Research On DFFD And Neural Network Model In Aerodynamic Optimization Of Cascade

The development and manufacturing plans for high-performance aeroengines are very important in the technological development plans of countries around the world.The performance of the compressor often has a direct impact on the aeroengine.Therefore,high-performance aeroengines place more demanding technical requirements on compressors.In order to improve the aerodynamic performance of compressors,domestic and foreign researchers have carried out a lot of research work on the optimization of design solutions.Firstly,this paper systematically investigates the CFD aerodynamic optimization design method,cascade parameterization method,Design of Experiments method,and objective function establishment and fault-finding method involved in the aerodynamic optimization design method for compressor cascade,and constructs a set of aerodynamic optimization design method.Then,different bivariate multipeak functions were selected.The LHS method is used to collect sample points to approximate and compare the functions for BPNN and RBFNN.The Laval nozzle case and the sod shock tube case were selected as the benchmark cases for solving the multilevel mesh,and verification and validation work was completed.The RAE2822 airfoil is chosen for aerodynamic optimization design.The results show that the drag coefficients of the two optimized airfoil are reduced by 22.4% and 26.3%,respectively.After the application of the methods involved in the optimization design,the aerodynamic optimization design with the total pressure loss after the cascade as the optimization objective was carried out with the low-speed controllable diffusion cascade as the object of study.The verification and validation work of the CFD calculation results of the low-speed controllable cascade were completed.,and the cascade was parameterized using DFFD.After training BPNN and RBFNN using numerical calculation results and applying genetic algorithm for optimization,the cascade flow field before and after optimization was compared and analyzed.The analysis of the optimization results shows that the total pressure loss is reduced by 7.77% after the two-dimensional cascade optimization,but the optimized cascade does not provide a better Direct application to three-dimensional cascade.The calculation results of the optimized leaf pattern for three-dimensional cascade show that the total pressure loss after the optimized cascade of BPNN and RBFNN,respectively,is reduced 2.52 percent and 3.20 percent,both of which improved mobility considerably.