NIR汽油辛烷值测定仪中的支持向量机方法  

Support Vector Machines for a NIR Spectroscopy Gasoline Octane Number Analyzer

作　　者：杜树新[1] 吴铁军[1] 

机构地区：[1]浙江大学工业控制技术国家重点实验室

出　　处：《仪器仪表学报》2004年第5期582-586,共5页Chinese Journal of Scientific Instrument

摘　　要：提出了应用于近红外光谱汽油辛烷值测定仪的支持向量机方法 ,该方法具有强的泛化能力及全局最优解的特点 ,得到的数学模型其预测能力明显改善。A novel algorithm applied to near-infrared (NIR) spectroscopy gasoline octane number analyzers is proposed, based on support vector machines (SVMs). Owing to the method's high generalization performance and global optimal solution, the obtained mathematic model shows obvious improvement in prediction ability. The experimental results show the proposed method has evident advantage over the classical algorithms such as PLS and ANN.

关 键 词：常见 NIR 近红外光谱 方法 测定仪 实验 偏最小二乘 支持向量机 泛化能力 人工神经网络 

分 类 号：TH744.4[机械工程—光学工程] TP183[机械工程—仪器科学与技术]