卷烟焦油预测研究  

Cigarette tar delivery prediction

作　　者：梁德成[1] 王德吉[1,2,3] 邱道尹[1] 栗卫军[2] 

机构地区：[1]华北水利水电学院电力学院,郑州450011 [2]中国烟草总公司职工技术培训中心,郑州450008 [3]中国科学院合肥智能所,合肥230031

出　　处：《东南大学学报（自然科学版）》2009年第S1期195-198,共4页Journal of Southeast University：Natural Science Edition

基　　金：河南省自然科学基金资助项目(200510078009)

摘　　要：为了解决传统预测方法不能准确预测焦油量的一小样本非线性预测问题,提出从支持向量机的角度进行预测,并在其中关键问题核函数的选择上,应用保角映射进行核函数构造.该算法将焦油预测问题通过非线性变换映射到一个高维的特征空间,在高维的特征空间构造线性判别函数以替换原空间的非线性判别函数.选择22种不同地区和等级的烟叶,利用相同卷烟纸和滤嘴将各种烟叶卷制成单料卷烟,检测得各种烟叶单克烟丝烟雾中焦油量,以此为样本,进行函数训练并进行预测.实验证明该算法能提高焦油量预测精度和效率.Prediction of tar delivery in cigarette smoke is difficult by traditional methods due to the very limited number of samples and the nonlinear relationship among variables.A SVM(support vector machine) forecast method is therefore proposed.The selection of the kernel,which function,which is the key point of the method,is based on the conformal mapping method.The method maps the prediction into a high dimensional eigen space via non-linear mapping.A linear function is constructed in the high dimensional eigen space and the original non-linear function is hence replaced.A total of 22 different type and grade of leaves are chosen to make single material cigarette and the sample for training the function is obtained.The function is tested thereafter.Results show that the algorithm can predict tar delivery better than the current method used.

关 键 词：支持向量机 卷烟焦油量 预测 

分 类 号：TS411[农业科学—烟草工业]