基于SVR模型预测超薄浮法玻璃气泡数量的研究  

Prediction of the Number of Bubbles in Ultra-thin Float Glass Based on SVR Model

作　　者：胡睿 田英良[1,2] 李小青 赵志永 韩正伟 李聚锋 李彦涛 HU Rui;TIAN Yingliang;LI Xiaoqing;ZHAO Zhiyong;HAN Zhengwei;LI Jufeng;LI Yantao(Faculty of Materials and Manufacturing,Beijing University of Technology,Beijing 100124,China;National Engineering Laboratory for Industrial Big-data Application Technology,Beijing University of Technology,Beijing 100124,China;Hebei Panel Glass Co.,Ltd,Yongqing 065600,China)

机构地区：[1]北京工业大学材料与制造学部,北京100124 [2]北京工业大数据应用技术国家工程实验室,北京100124 [3]河北视窗玻璃有限公司,河北永清065600

出　　处：《玻璃搪瓷与眼镜》2021年第6期1-6,共6页Glass Enamel & Ophthalmic Optics

基　　金：国家重点研发计划(2016YFB0303700)。

摘　　要：针对超薄浮法玻璃生产过程存在反馈滞后大、生产复杂、产品质量难以精确控制的情况以及气泡质量缺陷严重影响产品应用的问题,本文采用SVR模型对超薄浮法玻璃气泡数量进行预测,结合实际生产数据对模型进行验证,并探究熔化时滞时间以及特征属性对模型精度的影响,采用Relief-F算法结合实际建模结果筛选出对气泡产生有影响的重要特征属性。研究结果表明,SVR模型能精确预测玻璃气泡数量,为后续生产提供决策支持;熔窑池底温度与熔窑碹顶温度对超薄浮法玻璃生产中气泡的产生有较大影响。Due to the many problems in the production of ultra-thin float glass,such as long feedback delay,complicated process,difficult control of product quality and troublesome bubble defect which has serious influence on the application of the product,the SVR model is established to predict the number of bubbles in ultra-thin float glass and its verification has been carried out by actual production data.The effect of time lag and characteristic attributes of melting process on the accuracy of the model was investigated.The Relief-F algorithm combined with the actual modeling results is applied to sift out the important characteristic attributes that have an impact on the bubble generation.The results show that the SVR model is capabale to accurately predict the number of glass bubbles and provide decision support for subsequent production.The temperature of the bottom and the crown of the glass melting furnace have a greater impact on the generation of bubbles in the production of ultra-thin float glass.

关 键 词：超薄浮法玻璃 气泡数量 预测 支持向量回归 Relief-F算法 

分 类 号：TQ171.65[化学工程—玻璃工业]