基于核偏最小二乘的砷盐净化除钴过程钴离子浓度软测量  

Soft sensor based on kernel partial least squares for cobalt ion concentration in cobalt removal purification process with arsenic salt

作　　者：王雅琳[1] 黄凯华[1] 伍铁斌[1] 谢文平[1] 阳春华[1] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《中南大学学报（自然科学版）》2015年第1期141-148,共8页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(61273187);国家科技支撑计划项目(2012BAF03B05);高等学校博士学科点专项科研基金优先发展领域课题(20110162130011);国家创新研究群体科学基金资助项目(61321003)~~

摘　　要：针对锌湿法冶炼砷盐净化除钴过程中钴离子浓度无法在线检测给生产优化控制带来困难的问题,建立基于机理模型和核偏最小二乘(kernel partial least squares,KPLS)参数辨识的钴离子浓度软测量模型;考虑到过程具有时变性,根据所建立的软测量模型特点,提出一种双向递归KPLS模型参数更新和滤波修正相结合的模型在线校正方法,以提高软测量模型精度;同时,采用基于主元分析和贝叶斯分类的异常值在线检测方法实现对参数辨识相关检测量的实时异常值在线检测,保证用于参数更新数据的有效性。研究结果表明:所建钴离子浓度软测量模型跟踪效果好,满足实际生产过程预测精度要求,解决了钴离子浓度无法在线检测给优化控制带来的困难,可为生产过程的优化控制提供指导。Considering that the cobalt ion concentration in the cobalt removal purification process with arsenic salt can not be measured online, which makes it difficult to control and optimize the production process, a soft sensor model based on the mechanism and kernel partial least squares was built to predict cobalt ion concentration online. Considering the time-varying characteristics of the process, an online correction method was proposed to improve the precision of the soft sensor model by combining the bidirectional recursive KPLS model parameters update with the filtering correction. Meanwhile, in order to ensure the effectiveness of data samples which were used to model update, an online outlier detection method based on principal component analysis and Bayesian classification was adopted. The results show that the soft sensor model for cobalt ion concentration has good tracking performance. It meets the detection requirement for on-site process and solves the problem of optimal control, so it can provide effective operation guidance for control and optimization of the industrial process.

关 键 词：砷盐净化除钴过程 核偏最小二乘 软测量 模型在线校正 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]