摘要
对超细晶粒钢热模拟试验工艺参数:化学成分、变形温度、变形量以及变形道次等与超细晶铁素体晶粒尺寸及质量分数之间建立了映射关系。在此基础上建立了相应的超细晶粒钢组织预测模型。应用遗传算法对该神经网络模型的权值进行优化,从而克服了神经网络训练速度慢、容易陷入局域极小和全局搜索能力弱等缺点,提高了神经网络的预测精度。通过实例验证表明,超细晶铁素体晶粒尺寸的预测精度达90%以上,超细晶铁素体晶粒质量分数预测精度达91%以上。
The mapping relationship between thermal simulation technique parameters, such as chemical component, deformation temperature, deformation quantity and deformation times, and F fine dimension and content F fine in ultrafine grain steel was set up. The microstructure prediction model of the ultrafine grain steel was constructed. Genetic algorithm(GA) was used to optimize the weights of the artificial neural network(ANN) model to overcome some ANN weaknesses such as the slow training speed, easy to be plung...
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2006年第11期152-155,共4页
Transactions of the Chinese Society for Agricultural Machinery
基金
江苏省高校自然科学基金重点资助项目(项目编号:04KJA430021)
关键词
超细晶粒钢
组织预测
神经网络
遗传算法
Ultrafine grain steel
Microstructure prediction
Neural network
Genetic algorithm
