Compressor Blades Pneumatic-based Neural Network And Genetic Algorithm To Optimize The Design Study

The most important part of the aircraft engine is the compressor, which is also the hindrance of the aircraft engine development. In order to improve the population performance of the engine, it is necessary to enhance the performance of the compressor. The blade is the basic part of the compressor, also its geometry and aerodynamic performance affect the engine greatly. The experts of thermal-gasdynamics in turbo- machinery focus on the theory and design at all times, because the design of compressor is the key step. The optimization design of the cascade will be studied in this paper.Genetic algorithm, nonnumeric global optimization algorithm, is used in the aerodynamic design of the cascade design, and elitist model, niche model, pseudo-parallel genetic algorithms model are also introduced. The results show that pseudo-parallel genetic algorithms model takes on better rate of the convergence and ability of optimization. So pseudo-parallel genetic algorithms model will be the dominating algorithm.For the cost of the calculation, many variations and the analyses of aerodynamic performance, it is difficult to introduce genetic algorithm into the auto optimization design of the compressor blades. The analyzes step of genetic algorithm adaptation is added to the artificial neural network regression operation, then the aerodynamic optimization model of the compressor blades is established, at the end the system optimization rate will be accelerated with the fast mapping ability of artificial neural network.Combined with parameterization method of cascade based on non-uniformed B_spline, genetic algorithms, technology of artificial neural network, numerical simulation of Navier-Stokes equation and method of structured girds, a general and extendable axial compressor cascades aerodynamic optimization design system is established, the two-dimensional cascades profile aerodynamic direct optimize design is carried out for one type of cascade which stress pervasion is under control, which also verified the optimization system.