摘要
对中红外光谱快速鉴别新润滑油和废润滑油的方法进行了研究。收集了116个新润滑油样和32个废润滑油样,并随机抽取其中的81个新润滑油样和18个废润滑油样作为建模集,剩余的油样作为预测集。采用主成分分析法对建模集的中红外光谱数据进行变量筛选,建立了偏最小二乘法快速鉴别模型。该模型预测集的相关系数(r)和预测误差均方根(RMSEP)分别为0. 94445和0. 16551,鉴别率为98%。说明该方法具有很好的预测效果。
The identification method of new lubricants and used lubricants by mid-infrared spectroscopy was studied. 116 new lubricating oil samples and 32 used lubricating oil samples were collected,and 81 new lubricating oil samples and 18 used lubricating oil samples in them were randomly selected as the modeling set. The remaining samples were used as the prediction set. Principal component analysis( PCA) was used to select the effective infrared spectra data,and combined with partial least square( PLS) to build model for identification of new lubricants and used lubricants. The correlation coefficient( r) and root mean square error of prediction( RMSEP)were 0. 94445 and 0. 16551 respectively,and the discrimination rate reached 98%. The result showed that the method has good prediction effect.
作者
段小娟
蔡发
黄杰
邓可
戚佳琳
DUAN Xiao-juan;CAI Fa;HUANG Jie;DENG Ke;QI Jia-lin(Qingdao Entry-exit Inspection and Quarantine Bureau,Qingdao 266001,China;Technical Center of Shandong Entry-exit Inspection and Quarantine Bureau,Qingdao 266002,China)
出处
《润滑油》
CAS
2018年第5期55-57,64,共4页
Lubricating Oil
关键词
中红外光谱
润滑油
主成分分析
偏最小二乘法
mid- infrared spectroscopy
lubricating oil
principal component analysis
partial least square