摘要
用神经网络建模,遗传算法优化是求解工程优化问题的一种行之有效的方法.但由于神经网络、遗传算法自身的诸多问题,制约了其在较为复杂领域中的应用.正交试验法是一种进行多因素试验的科学方法,本文将它引入神经网络和遗传算法中,将三者结合起来,用正交试验法设计神经网络样本,并用遗传算法优化神经网络结构,较好地克服了各自的弱点,拓展了它们的应用.另外本文对正交试验法配置遗传算法参数做了介绍.
It is an efficient method to solve engineering optimization problems with neural networks and genetic algorithms.Because of many defects in themselves,the applications of neural networks and genetic algorithms are restricted and it's almost impossible to use them in complicated fields.Orthogonal experimental method is a scientific experimental method of many factors.This paper introduces this method,combined with neural networks and genetic algorithms.In this way we overcome their respective defects to some degree and broaden their applications.
出处
《南昌大学学报（工科版）》
CAS
2003年第1期79-82,共4页
Journal of Nanchang University(Engineering & Technology)
关键词
正交试验法
神经网络
遗传算法
铝型材挤压
orthogonal experimental method
neural networks
genetic algorithms
extrusion
