基于实验、神经网络和遗传算法的一种优化设计方法

An Optimal Design Based on Experiment, Neural Networks and Genetic Algorithm

针对传统优化设计方法对于机理复杂情况难以给出明确函数及只能寻优局部极值的缺陷 ,提出了用实验数据和神经网络形成“虚拟”函数和用遗传算法寻优全局极值的一种优化设计方法 。

It is difficult for the traditional optimization to get the function between the objects and the variables in the complex case. With this traditional method the better solution can only been found in local area. The paper describes an optimization method, which produces a virtual function between the objects and the variables according to the experiment data and artificial neutral networks learning at first, then finds the best answer using genetic algorithm. It also gives an example.