摘要
提出一种基于最小二乘支持向量机(LS SVM)的柴油十六烷值近红外光谱测量方法。该方法用聚类分析法对训练样本进行了一次筛选,然后用最小二乘支持向量机建立柴油十六烷值的预测模型。实验结果表明该方法不仅可以显著减少计算时间,在预测精度上比常用的多元线性回归和偏最小二乘等方法有显著提高。
A novel method applied to near-infrared(NIR)spectroscopy diesel cetane number prediction is presents.The method is based on least square support vector machine(LS-SVM).The prediction model is built on a training set selected by clustering method.Experiment results show that the proposed method has evident advantage over the classical methods such as multi-variant linear regression and partial least square.
出处
《化工自动化及仪表》
CAS
北大核心
2004年第2期48-51,共4页
Control and Instruments in Chemical Industry
关键词
最小二乘支持向量机
聚类分析
近红外光谱
柴油十六烷值
least square support vector machine
cluster analysis
near-infrared spectroscopy
diesel cetane number
