东北大学学报:自然科学版 ›› 2015, Vol. 36 ›› Issue (10): 1408-1411.DOI: 10.3969/j.issn.1005-3026.2015.10.009

• 信息与控制 • 上一篇    下一篇

基于潜变量模型的复杂过程生产设计方法

王小刚1, 沙毅1, 朱丽春2   

  1. (1. 东北大学 信息科学与工程学院, 辽宁 沈阳110819; 2. 中国科学院 国家天文台, 北京100012)
  • 收稿日期:2014-09-09 修回日期:2014-09-09 出版日期:2015-10-15 发布日期:2015-09-29
  • 通讯作者: 王小刚
  • 作者简介:王小刚(1960-),男,辽宁沈阳人,东北大学副教授.
  • 基金资助:
    国家自然科学基金资助项目(11273001).

Production Design Method Based on Latent Variable Model for Industry Processes

WANG Xiao-gang1, SHA Yi1, ZHU Li-chun2   

  1. 1.School of Information Science & Engineering, Northeastern University, Shenyang 110819,China; 2.National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, China.
  • Received:2014-09-09 Revised:2014-09-09 Online:2015-10-15 Published:2015-09-29
  • Contact: WANG Xiao-gang
  • About author:-
  • Supported by:
    -

摘要: 针对工业品生产过程控制中输入变量的确定及零空间问题求解等复杂过程的生产设计问题,在现有的多元潜变量建模及生产设计等相关问题研究成果的基础上,深入研究了主元回归方法的应用策略和生产设计问题,提出了一种基于潜变量模型的复杂过程生产设计方法.通过与基于标准回归模型生产设计方法对比,展示了该方法的优越性.该生产设计方法的操作条件不仅能够满足工业生产过程产品质量的要求,而且与历史工况的关联结构和范围保持一致,为零空间问题的解决提供了可行方案.最后通过仿真验证了该类方法求解含有零空间的生产设计问题的有效性.

关键词: 生产设计, 主元回归, 零空间, 多元潜变量, 模型

Abstract: In view of determining the manipulated variables in the industrial control process and solving the null space problems during the production design of the complex industrial process, an in-depth study about the application strategy of the principal component regression was done on the basis of the existing multi-latent variable modeling methods. In addition, a production design strategy based on the latent variable was proposed. Compared with the standard regression model, the advantages of the multivariate latent variable method on solving production design problems were revealed. The proposed design method could not only satisfy the requirement of product quality, but also be consistent with the correlation structure of the historical condition and scope, which could be regarded as a feasible scheme to solve the null space problems. The simulation results showed the effectiveness of the designed method for solving production design problems of null space.

Key words: product design, principal component regression, null space, multivariate latent variable, model

中图分类号: