上海交通大学学报(自然版) ›› 2012, Vol. 46 ›› Issue (12): 1951-1955.

• 能源与动力工程 • 上一篇    下一篇

基于模型迁移的磨矿过程混合蛙跳算法-小波神经网络软测量建模及重构

王介生,杨阳,孙世峰   

  1. (辽宁科技大学 电子与信息工程学院, 辽宁 鞍山 114044)  
  • 收稿日期:2012-05-25 出版日期:2012-12-29 发布日期:2012-12-29
  • 基金资助:

    中国博士后科学基金项目(20110491510),辽宁省高等学校优秀人才支持计划(LJQ2011027)

     

SFLA-WNN Soft-Sensor Modeling and Reconfiguration of Grinding Process Based on Model Migration

 WANG  Jie-Sheng, YANG  Yang, SUN  Shi-Feng   

  1. (School of Electronic and Information Engineering, University of Science & Technology Liaoning, Anshan 114044, China)
  • Received:2012-05-25 Online:2012-12-29 Published:2012-12-29

摘要: 摘要: 
以磨矿过程的关键工艺指标(磨矿粒度和磨机排矿速率)为预测对象,提出一种基于小波神经网络(WNN)的自适应软测量建模方法.通过对磨矿过程工艺的分析,选取了软测量模型的辅助变量,利用混合蛙跳算法(SFLA)对WNN软测量模型的结构参数(小波函数伸缩因子、平移因子和网络连接权重)进行优化,实现软测量模型输入输出变量之间的非线性映射;并采用模型迁移思想及输入输出修正规划方法实现软测量模型的重构,以解决输入矿石品位改变这一动态工况下的模型自适应校正问题.仿真结果表明,所提出的模型能够显著提高磨矿过程中经济技术指标预测的精度和鲁棒性,满足磨矿生产过程的实时控制要求.
关键词: 
磨矿过程; 软测量; 小波神经网络; 混合蛙跳算法; 模型迁移
中图分类号:  TK 232
文献标志码:  A    

Abstract: For forecasting the key technology indicators (grinding granularity and mill discharge velocity) of grinding process, an adaptive soft-sensor modeling method based on wavelet neural network (WNN) was proposed. The assistant variables of the soft-sensor model are selected by analyzing the technique characteristic of the grinding process. The structure parameters (scaling factors and translation factors of the wavelet functions, connections weights) of the WNN are optimized by the shuffled frog leaping algorithm (SFLA) to realize the nonlinear mapping between input and output variables of the discussed softsensor model. Model migration strategy and input-output space bias correction (IOSBC) method are adopted to realize the on-line adaptive revision of soft-sensor model. The simulation results show that the proposed model can significantly enhance the predictive accuracy and robustness of the technicaland-economic indexes and satisfy the real-time control requirements of the grinding process. Key words:

Key words: grinding process, soft-sensor, wavelet neural network (WNN), shuffled frog leaping algorithm(SFLA), model migration