基于并行遗传算法压气机叶片自动优化设计

结合小生境法与算子自适应法对基本遗传算法进行改进,采用WinSock接口、多线程、CS体系结构,实现并行遗传算法.将并行遗传算法与NS方程流场计算方法、Hicks-Henne函数叶型参数化方法结合构成叶型自动优化设计软件,软件可用于叶型设计和对已有叶型的改进.在构造目标函数时,根据工作攻角范围,构造了考虑非设计点性能的目标函数.分别对进口马赫数为0.5和0.9左右的叶栅采用三台计算机进行并行气动优化设计,优化叶型设计点和非设点性能都明显好于原始叶型.

Design optimization of compressor blades using parallel genetic algorithms

Simple Genetic Algorithm is improved with Niche and adaptive operator,the new algorithm called Improved Genetic Algorithm(IGA) is parallelized using WinSock,multithread and Client-Server architecture.Blade profiles are parameterized using Hicks Henne function,2D cascade flow fields are calculated by using NavierStokes equation method.To form software for design optimization of compressor blade profiles,the IGA is combined with the above blade profile parameterization and flow field calculation method.The software can be used for design of compressor blades or improvement of existing blades.In this paper,design of two compressor blade profiles is taken as examples.The inlet Mach number of two cascades composed of the blades is about 0.5 and 0.9,respectively.To insure improvement of cascade aerodynamic performances in whole work ranges,the objective functions are constructed from consideration of both design and off design point performances.In the optimization procedure,three computers are run in parallel to reduce design time.The aerodynamic performances of optimal blades are obviously better than that of the initial blades.