【作者】 陈恒志； 杨建平； 余相灼； 刘青；

【Author】 Chen Hengzhi;Yang Jianping;Yu Xiangzhuo;Liu Qing;State Key Laboratory of Advanced Metallurgy, University of Science and Technology Beijing;

【机构】 北京科技大学钢铁冶金新技术国家重点实验室；

【摘要】 为提高连铸坯质量预测模型的预测精度,提出了将主成分分析与GA-BP神经网络相结合的连铸坯质量预测方法。使用主成分分析法对多个影响连铸坯质量的因素进行降维处理或重新组合,将处理后所得较少的主成分变量作为样本输入GA-BP神经网络进行训练而得到连铸坯质量预测模型,对方大特钢的60Si2Mn连铸坯中心疏松和中心偏析缺陷进行预测,并与以未经处理的影响因素作为输入变量的GA-BP神经网络连铸坯质量预测模型进行对比分析。结果表明:基于主成分分析与GA-BP神经网络相结合的连铸坯质量模型的预测精度较高,对连铸坯中心疏松和中心偏析缺陷的预测准确率分别为85%、80%,且模型的运算速度有了显著的提升。更多还原

【Abstract】 In order to improve the prediction accuracy of the quality prediction model of continuous casting bloom,a data processing method based on the combination of principal component analysis and BP neural network was presented. By using principal component analysis,the amount of variables which affect the quality of continuous casting will be reduced. Then the principal components are employed to train the BP neural network in order to obtain the quality prediction model of continuous casting bloom. As the number of inputs is reduced, the train process can be faster and the iteration time can be reduced. In comparison with the network model which uses the original variables as the inputs to predict the degree of the center porosity and central segregation of 60 Si2Mn continuous casting bloom produced by Fangda Special Steel, the results show that GA-BP neural network prediction modeling method based on principal component analysis method is improved to 85% and 80% in the center loose and central segregation defects respectively, and this modeling method has both high efficiency in calculation.更多还原