基于支持向量机的焦炭质量预测模型  

Prediction Model of Coke Quality Based on Support Vector Machines

作　　者：崔庆安[1] 何桢[1] 崔富新[2] 

机构地区：[1]天津大学管理学院,天津300072 [2]河南安阳钢铁集团公司,河南安阳455004

出　　处：《化工自动化及仪表》2006年第1期28-31,共4页Control and Instruments in Chemical Industry

基　　金：天津市科技攻关项目资助(04310881R)

摘　　要：采用机器学习方法中的支持向量机技术来预测焦炭质量,不需要了解煤成焦机理,综合考虑配合煤特性和焦炉加热制度的影响,而且克服了人工神经网络预测精度低的缺点,在取得最小拟合误差的同时可以得到最小的预测误差;给出了一种基于组态王6.0与VB6.0的焦炭质量在线预测系统框架。The support vector machines technology in machine learning field is adopted to predict the coke quality. Effects of coal blending indexes and coke oven heating indexes are considered comprehensively in the proposed method without understand the mechanism of coke formation. The defect of the low precision in the artificial intelligence neural networks based coke quality prediction is overcome. The minimum error of fitting and the minimum error of predicting are obtained in the meantime. A framework of online predicting system for coke quality based on CON6.0 and VB6.0 is also presented.

关 键 词：支持向量机 人工神经网络 机器学习 焦炭质量 在线预测 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]