基于遗传神经网络的不锈钢带冷轧轧制力模型  

Cold Rolling Force Model Based on GA and ANN for Stainless Steel Strip

作　　者：张清东[1] 徐兴刚[1] 于孟[1] 瞿标[2] 李实[2] 

机构地区：[1]北京科技大学机械工程学院,北京100083 [2]宁波宝新不锈钢有限公司,浙江宁波315807

出　　处：《钢铁》2008年第12期46-48,57,共4页Iron and Steel

摘　　要：为了提高工厂从国外引进的以Bland-Ford公式为基础的冷轧不锈钢带轧制力模型的计算精度,将基于遗传算法的BP神经网络与现有变形阻力和轧制压力解析数学模型相结合,建立了变形阻力和轧制压力修正模型。将在生产现场采集的部分过程记录数据,进行分类和预处理后作为训练样本用于训练遗传神经网络模型。将其他现场实测数据用于验证所建的轧制力模型,计算结果表明所建的轧制力模型具有较高的计算精度。In order to improve the precision of imported rolling force model and based on Bland-Ford formula, the modified yield stress model and rolling force model were established by combining mathematical models of yield stress and rolling force with BP neural network and Genetic Algorithm. Lots of actual measured data pretreated and divided according to steel grade were used to train the neural network. The comparison between the measured and predicted rolling force showed that the new rolling force model had higher prediction accuracy.

关 键 词：不锈钢 轧制力 神经网络 遗传算法 

分 类 号：TG335.12[金属学及工艺—金属压力加工]