English

应用科学学报 ›› 2005, Vol. 23 ›› Issue (6): 648-653.

• 论文 • 上一篇    下一篇

(Zr0.7 Sn0.3) TiO4陶瓷性能预报的支持向量回归模型

翟文华1, 陆文聪1, 刘旭1, 陈念贻1, 王国庆2   

  1. 1. 上海大学理学院, 上海 200444;
    2. 天津大学电子信息工程学院, 天津 300072
  • 收稿日期:2004-06-30 修回日期:2004-09-13 出版日期:2005-11-30 发布日期:2005-11-30
  • 作者简介:翟文华(1980-),男,山东威武人,硕士,E-mail:gerwena@163.com;陆文聪(1964-),男,浙江慈溪人,教授,博导,E-mail:wclu@staff.shu.edu.cn
  • 基金资助:
    国家自然科学基金(20373040)资助项目;上海市纳米专项(0252nm101)项目

Prediction of Properties of (Zr0.7Sn0.3)TiO4 Dielectric Ceramics Using Support Vector Regression

ZHAI Wen-hua1, LU Wen-cong1, LIU Xu1, CHEN Nian-yi1, WANG Guo-qing2   

  1. 1. School of Sciences, Shanghai University, Shanghai 200444, China;
    2. College of Electronics and Information Engineering, Tianjin University, Tianjin 300072, China
  • Received:2004-06-30 Revised:2004-09-13 Online:2005-11-30 Published:2005-11-30

PDF

77

摘要： 将支持向量回归方法用于(Zr0.7Sn0.3) TiO4陶瓷的配方性能关系研究中,分别建立了(Zr0.7 Sn0.3) TiO4陶瓷介电常数和损耗角正切的支持向量回归模型,并与逆传播人工神经网络、多元线性回归模型进行了比较.用留一法分别检验了支持向量回归、逆传播人工神经网络和多元线性回归3种不同模型的预报能力,结果表明:上述3种模型对于(Zr0.7Sn0.3) TiO4陶瓷介电常数的留一法预报的平均相对误差分别为1.083%、1.632%、1.931%,对于损耗角正切的留一法预报的平均相对误差分别为0.999%、1.746%、1.414%.因此,支持向量回归模型的预报能力较好,可望在陶瓷配方设计中的多变量、非线性问题和小样本体系中发挥较好的作用,为新型介电陶瓷的性能预报和配方优化提供一条全新可靠的途径.

关键词: (Zr0.7Sn0.3) TiO4, 介电性能, 留一法, 支持向量机, 支持向量回归

Abstract: The support vector regression (SVR) is applied to investigate the relationship between the properties and formulation of (Zr0.7Sn0.3) TiO4 dielectric ceramic.The models for predicting dielectric constant and loss angle tangent are proposed respectively.Using the leave-one-out cross-validation method, the mean relative errors (MRE) of dielectric constant in SVR, back-propagation artificial neural network (BP-ANN) and multiple linear regression (MLR) models are checked, with the results of 1.083%, 1.632% and 1.931% respectively.Similarly, the MRE of loss angle tangent in SVR, BP-ANN and MLR are 0.999%, 1.746% and 1.414% respectively.It is therefore demonstrated that support vector regression is a useful tool in dealing with multi-variable problems with non-linearity or small size of data set in the formulation design of dielectric ceramics.

Key words: (Zr0.7 Sn0.3) TiO4, dielectric properties, support vector machine, support vector regression, le ave-one-out

中图分类号: