一个神经网络结合遗传算法的叶轮逆命题设计方法  

Impeller Inverse Design by Neural Networks Coupled with Genetic Algorithms

作　　者：樊会元[1] 席光[1] 王尚锦 

机构地区：[1]西安交通大学能源与动力工程学院机泵中心,陕西西安710049

出　　处：《航空动力学报》2000年第1期47-50,共4页Journal of Aerospace Power

摘　　要：本文的目的是利用神经网络实现离心压缩机叶轮的逆命题设计。为此,结合实例叶轮逆命题设计问题通过遗传算法的演化和训练建立了前馈神经网络。所建网络随之用于求解给定表面速度分布情况下的叶片型线。数值模拟显示。A study on application of neural networks to solving an inverse design of a centrifugal compressor impeller has been conducted.The feedforward neural networks were established by means of evolving and training with genetic algorithms for a real impeller inverse design,and then used to predict the blade profile when the velocity distribution on its surface is given.Numerical simulation shows that the neural networks can successfully yield the desired blade profile.

关 键 词：神经网络 离心式压缩机 叶轮 设计 遗传算法 

分 类 号：TH452[机械工程—机械制造及自动化] TB115[理学—数学]