基于最大互信息系数属性选择的冷轧产品机械性能预测

来源期刊:中南大学学报(自然科学版)2020年第1期

论文作者:吕志民 颜弋凡 安路达

文章页码:68 - 76

关键词:最大互信息系数;冷轧;机械性能预测;属性选择

Key words:maximal information coefficient; cold rolling; mechanical property prediction; feature selection

摘    要:对于冷轧产品机械性能预测建模时面对的需要对全流程众多影响工艺参数属性选择的问题,提出基于最大互信息系数(MIC)属性选择的机械性能预测建模方法。该方法首先利用MIC算法计算各性能指标和工艺参数之间的相关性度量,然后根据各相关度量选择形成工艺参数属性子集用于性能预测模型建模及预测。研究结果表明:该建模方法构建的冷轧产品性能预测模型的预测精度高于全工艺参数模型、Pearson相关系数选择和经验知识选择,另外该方法也能选择出一些传统方法不能选择出的非线性影响关系的工艺参数。最优特征子集模型预测效果从原始全工艺参数模型的平均相对误差2.90%下降到2.30%。

Abstract: The selection of process parameters in the throughout manufacturing process influenced the accuracy of prediction model of the mechanical properties of cold rolled products. A mechanical properties prediction modeling method was proposed based on the maximal information coefficient (MIC) attribute selection. Firstly, the MIC algorithm was used to calculate the correlation metric between each mechanical property index and process parameters, and then the relevant metrics ware selected to form a subset of process parameter attributes for mechanical properties prediction modelling. The results show that the prediction accuracy of the cold rolling product properties prediction model constructed by the modeling method is higher than the full process parameter model, Pearson correlation coefficient selection and empirical knowledge selection. In addition, the method can also select some process parameters with nonlinear influence relationships between mechanical property indexes that cannot be selected by the traditional method. The optimal feature subset model prediction effect decreases from the mean relative error of the original full process parameter model by 2.90% to 2.30%.

相关论文

  • 暂无!

相关知识点

  • 暂无!

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号