摘要
提出了一种基于BP神经网络和遗传算法(GA)的多工况离散变量结构优化设计方法,并对某斗轮堆取料机回转平台进行优化设计。该方法将多工况问题处理为多约束问题,利用正交试验法选择神经网络训练样本点,通过参数化有限元模型计算出各工况下的样本数据,建立起基于BP神经网络的回转平台数学模型,为遗传算法提供适应度函数,最后运用遗传算法完成寻优计算。结果表明,回转平台自重减轻13.8%,取得了满意的优化效果。
A method of structure optimization for discrete variables under multiple load cases was proposed based on the BP neural network and genetic algorithm（GA）.Optimal design for slewing platform of bucket stacking-reclaiming machines was carried out by the method.Translating multiple load cases into multiple constraints,utilizing the Orthotropic Experimental Method（OME） to select the training sample points and calculating the sample data under every load case by the parameterized finite element model,a mathematical model of system was built on the basis of the BP neural network.Through optimizing the neural network by genetic algorithm,the results have proved that the weight of the slewing platform can be decreased 13.8%.
出处
《机械设计与研究》
CSCD
北大核心
2012年第1期105-108,120,共5页
Machine Design And Research
关键词
回转平台
结构优化
多工况
离散变量
神经网络
遗传算法
slewing platform
structure optimization
multiple load cases
discrete variables
neural networks
genetic algorithm
