遗传算法与神经网络协作求解结构优化问题

NEURAL NETWORKS COOPERATED WITH GENETIC ALGORITHM IN SOLVING OPTIMAL STRUCTURAL PROBLEMS

本文简要介绍了遗传算法(简称GA)求解问题的步骤以及解决实际问题时的优缺点.为了利用GA的优点,提高其收敛速度,本文提出GA与神经网络(BP网络)协作,利用神经网络的联想记忆、特征提取的能力,辅助GA求解结构优化设计问题,以避免在GA中所做的那些不必要的分析计算,从而节省了计算时间,最后通过算例证实,计算时间可减少15—20%.

This paper presents a method in which neural networks is combined with the genetic algorithm in solving optimal structural problems. The method uses the ability of associative memory and abstracting characteristics of neural networks (BP) to avoid unnecessary structural analyses, and retains the advantage of genetic algorithm and accelarates the process of convergence.