东北大学学报(自然科学版) ›› 2009, Vol. 30 ›› Issue (2): 172-175.DOI: -

• 论著 • 上一篇    下一篇

基于模糊基函数变换的PLS及其在软测量中的应用

贾润达;毛志忠;常玉清;   

  1. 东北大学信息科学与工程学院;东北大学流程工业综合自动化教育部重点实验室;
  • 收稿日期:2013-06-22 修回日期:2013-06-22 出版日期:2009-02-15 发布日期:2013-06-22
  • 通讯作者: Jia, R.-D.
  • 作者简介:-
  • 基金资助:
    国家高技术研究发展计划项目(2006AA060201)

Partial least squares regression based on transformation of fuzzy basis functions and its application to soft sensor

Jia, Run-Da (1); Mao, Zhi-Zhong (1); Chang, Yu-Qing (1)   

  1. (1) School of Information Science and Engineering, Northeastern University, Shenyang 110004, China; (2) Key Laboratory of Integrated Automation of Process Industry, Northeastern University, Shenyang 110004, China
  • Received:2013-06-22 Revised:2013-06-22 Online:2009-02-15 Published:2013-06-22
  • Contact: Jia, R.-D.
  • About author:-
  • Supported by:
    -

摘要: 针对以往非线性偏最小二乘法(PLS)的局限性,提出了一种基于模糊基函数变换的PLS算法,用以解决非线性系统的建模问题.该方法首先通过模糊基函数变换将自变量与因变量之间的非线性关系转化为线性关系,再利用PLS算法进行回归求参,从而有效克服由于模糊基函数变换所引发的维数增加以及多重共线性,使得模型在具有较高拟合精度的同时还能有效地抑制噪声.通过仿真实验进一步验证了该方法的有效性,并将其应用于湿法冶金萃取过程组分含量软测量建模问题,得到了满意的效果.

关键词: 模糊基函数, 非线性, 偏最小二乘法, 软测量, 萃取

Abstract: To model the nonlinear system and get rid of the limitation of existing nonlinear partial least squares (PLS), a nonlinear PLS algorithm based on the transformation of fuzzy basis functions is presented. The nonlinear relationship between independent and dependent variables are changed into linear one by transforming the fuzzy basis functions. Then, the PLS algorithm is used to get the regression parameters of the transformed independent and dependent variables so as to efficiently solve the problems of dimension increasing and multi-collinearity caused by the transformation of fuzzy basis functions. The model thus developed has the goodness of fit and is available to restrain the noise in process data. Simulation test results verified the superiority of this method to other nonlinear PLS methods, and it has been applied to the soft sensing modeling for component contents in hydrometallurgy extraction process with satisfactory prediction results obtained.

中图分类号: