基于实测数据的热轧带钢变形抗力模型
|
摘要
采用热连轧生产过程实测数据,建立一种新型结构的精轧带钢变形抗力模型。首先,提出热轧带钢变形抗力广义相加模型的建立方法,包括清洗数据、建立子模型、验证子模型、修正子模型等建模步骤,给出了求解广义相加模型的局部积分算法。接着,针对某热连轧生产线进行变形抗力建模实验,基于收集到的大量热连轧带钢生产历史数据,根据轧制机理和先验知识对各模型自变量进行函数变换预处理,再采用参数估计方法并指定变形抗力模型的基本形式,通过局部积分算法迭代计算确定各自变量的单变量光滑函数。预测实践表明,新模型计算精度高于该热连轧机组在线模型,具有计算精度高、适应性强等优点。
A deformation resistance model with new structure for finishing rolled strip is established based on the measured data of hot strip rolling process. Firstly,the establishment method of the generalized additive model of deformation resistance of hot rolled strip steel is proposed,including the modeling steps of cleaning data,establishing submodel,validating submodel,modifying submodel,and the local scoring algorithm for solving the generalized additive model is given. Then,aiming at prediction experiment of deformation resistance model of some hot-rolled strip production line,based on a large number of history data of steel strip production,the function transformation preprocessing of each model independent variable is carried out according to the rolling mechanism and prior knowledge,and then the parameter estimation method is adopted and the basic form of the deformation resistance model is specified,and the respective variable univariate smooth function is calculated iteratively through the local scoring algorithm. The prediction practice has shown that the calculation accuracy of the new model is higher than that of the online model of the hot rolling mill,and it has the advantages of high calculation accuracy and strong adaptability.
A deformation resistance model with new structure for finishing rolled strip is established based on the measured data of hot strip rolling process. Firstly,the establishment method of the generalized additive model of deformation resistance of hot rolled strip steel is proposed,including the modeling steps of cleaning data,establishing submodel,validating submodel,modifying submodel,and the local scoring algorithm for solving the generalized additive model is given. Then,aiming at prediction experiment of deformation resistance model of some hot-rolled strip production line,based on a large number of history data of steel strip production,the function transformation preprocessing of each model independent variable is carried out according to the rolling mechanism and prior knowledge,and then the parameter estimation method is adopted and the basic form of the deformation resistance model is specified,and the respective variable univariate smooth function is calculated iteratively through the local scoring algorithm. The prediction practice has shown that the calculation accuracy of the new model is higher than that of the online model of the hot rolling mill,and it has the advantages of high calculation accuracy and strong adaptability.
引文
[1]王健,王宇,谢红飙,等.基于热连轧实测数据的金属材料变形抗力模型[J].塑性工程学报,2015,22(1):55.
[2]李维刚,冯宁,赵云涛.基于广义相加模型的热轧变形抗力预测[J].钢铁研究学报,2018,30(6):447.
[3]王健,肖宏,郭振华.基于热轧生产实绩的变形抗力模型参数确立[J].武汉科技大学学报,2012,35(5):342.
[4]HASTIE T,TIBSHIRANI R. Generalized additive models[J]. Statistical Science,1986,1(42):297.
[5]JAMES G M,SILVERMAN B W. Functional adaptivemodel estimation[J]. Journal of the American StatisticalAssociation,2005,100(470):565.
[6]MULLER H G,YAO F. Functional additive models[J].Journal of the American Statistical Association,2008,103(484):1534.
[7]MCLEAN M W,HOOKER G,STAICU A M,et al. Func-tional generalized additive models[J]. Journal of Compu-tational and Graphical Statistics,2014,23(1):249.
[8]王圆圆,郭徵,李贵才,等.基于广义加性模型估算1979—2014年三峡库区降水及其特征分析[J].地理学报,2017,72(7):1207.
[9]王磊之,胡庆芳,王银堂,等.基于广义可加模型的降水空间估算模型[J].水利水运工程学报,2016(6):44.
[10]DREXLER M,AINSWORTH C H. Generalized additivemodels used to predict species abundance in the Gulf ofMexico:an ecosystem modeling tool[J]. Plos One,2013,8(5):e64458.
[11]HAN J,KAMBER M. Data mining:concepts and tech-niques second edition[M]. Burlington:Morgan Kauf-mann Pub.,2006.
[12]HASTIE T J,TIBSHIRANI R. Generalized additivemodels[J]. Statistical Science,1986,1(3):297.
[2]李维刚,冯宁,赵云涛.基于广义相加模型的热轧变形抗力预测[J].钢铁研究学报,2018,30(6):447.
[3]王健,肖宏,郭振华.基于热轧生产实绩的变形抗力模型参数确立[J].武汉科技大学学报,2012,35(5):342.
[4]HASTIE T,TIBSHIRANI R. Generalized additive models[J]. Statistical Science,1986,1(42):297.
[5]JAMES G M,SILVERMAN B W. Functional adaptivemodel estimation[J]. Journal of the American StatisticalAssociation,2005,100(470):565.
[6]MULLER H G,YAO F. Functional additive models[J].Journal of the American Statistical Association,2008,103(484):1534.
[7]MCLEAN M W,HOOKER G,STAICU A M,et al. Func-tional generalized additive models[J]. Journal of Compu-tational and Graphical Statistics,2014,23(1):249.
[8]王圆圆,郭徵,李贵才,等.基于广义加性模型估算1979—2014年三峡库区降水及其特征分析[J].地理学报,2017,72(7):1207.
[9]王磊之,胡庆芳,王银堂,等.基于广义可加模型的降水空间估算模型[J].水利水运工程学报,2016(6):44.
[10]DREXLER M,AINSWORTH C H. Generalized additivemodels used to predict species abundance in the Gulf ofMexico:an ecosystem modeling tool[J]. Plos One,2013,8(5):e64458.
[11]HAN J,KAMBER M. Data mining:concepts and tech-niques second edition[M]. Burlington:Morgan Kauf-mann Pub.,2006.
[12]HASTIE T J,TIBSHIRANI R. Generalized additivemodels[J]. Statistical Science,1986,1(3):297.