东北大学学报(自然科学版) ›› 2010, Vol. 31 ›› Issue (4): 494-497.DOI: -

• 论著 • 上一篇    下一篇

基于SVM和PSO的烧结工况预报方法

姜慧研;景世磊;柴天佑;周晓杰;   

  1. 东北大学软件学院;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:2013-06-20 修回日期:2013-06-20 出版日期:2010-04-15 发布日期:2013-06-20
  • 通讯作者: -
  • 作者简介:-
  • 基金资助:
    国家自然科学基金资助项目(60534010);;

Sintering condition prediction based on SVM and PSO

Jiang, Hui-Yan (1); Jing, Shi-Lei (1); Chai, Tian-You (2); Zhou, Xiao-Jie (2)   

  1. (1) School of Software, Northeastern University, Shenyang 110004, China; (2) Key Laboratory of Integrated Automation of Process Industry, Northeastern University, Shenyang 110004, China
  • Received:2013-06-20 Revised:2013-06-20 Online:2010-04-15 Published:2013-06-20
  • Contact: Jiang, H.-Y.
  • About author:-
  • Supported by:
    -

摘要: 利用滚球算法对在线采集的烧结工况图像进行去噪处理,然后利用大津方法和双快速行进法从去噪后的图像中分割出物料区、火焰区和充分燃烧区等关心区域,再从这些关心区域中提取特征.基于ReliefF-GA方法对特征进行约简,利用PSO优化SVM模型参数,建立烧结工况预报模型,基于该模型进行烧结工况预报.经过大量实验验证,该方法可以较大程度地提高烧结工况的预报率.

关键词: 烧结工况, 图像处理, 模式识别, 支持向量机, 粒子群算法

Abstract: Sintering condition images that were collected online in rotary kiln were denoised by rolling-ball algorithm, then the ROI (regions of interest) such as the material, flame and full combustion zones in the denoised images were segmented by Otsu method and double fast marching method. With some features extracted from the ROI then reduced on the basis of the RelieF-GA method, the SVM model parameters were optimized via PSO algorithm to develop a model to predict the sintering conditions. Lots of experimental results verified that this method can improve the prediction rate of sintering condition greatly.

中图分类号: