摘要
利用BP神经网络极强的非线性映射功能 ,成功地建立了封闭周转轮系的设计变量与模态柔度之间非线性映射的 3层BP神经网络模型 ,解决了动态优化设计时目标函数难以建立的难题 ,使非常复杂的动态优化问题变得十分简单。利用本文所提出的混合离散变量遗传算法对封闭周转轮系进行了动态优化设计 ,并对优化后的设计方案进行了动态特性分析 ,通过对比可以看出 ,优化后的设计方案不但动态特性有了较大提高 。
The intensive nonlinear mapping function of BP n eu ral net is applied successfully in establishing the nonlinear mapping three-lay er BP neural net relationship between the design variables and mode flexibility of closed epicyclic gear trains. So, the problem that objective function is hard to establish in dynamic optimization is solved, and this makes some complex dyn amic optimization very simple. A closed epicyclic gear train is dynamically opti mized by means of the genetic algorithm with mixing discrete variables. The dyna mic characteristics of the optimized design are analyzed. It is obvious that in the optimized design not only the dynamic characteristics are improved remarkabl y but also its weight is reduced.
出处
《机械科学与技术》
CSCD
北大核心
2004年第9期1131-1134,共4页
Mechanical Science and Technology for Aerospace Engineering
基金
江苏省教育厅自然科学基金项目 ( 0 2KGD460 0 8)资助
关键词
封闭周转轮系
动态优化设计
神经网络
遗传算法
Closed epicyclic gear trains
Dynamics optimizati on design
Neural net
Genetic algorithm
