摘要
针对近红外光谱分析技术中模型通用性较差的问题,提出了一种新的模型传递方法——最小角回归结合一元线性直接校正法(Least angle regression combined simple linear regression direct standardization,LARSLRDS)。该方法首先采用小波变换对样品光谱数据进行预处理,然后利用LAR实现样品全谱区光谱特征波长点的筛选,最后利用SLRDS对筛选出来的变量进行校正。采用汽油和药品样本的近红外光谱数据验证LAR-SLRDS性能,汽油数据集C7、C8、C9和C10成分的光谱差异为0. 002 8、0. 002 7、0. 002 6和0. 002 7,预测标准差为0. 410 6、0. 849 2、1. 034 9和1. 215 8;药品数据集活性、硬度和重量成分的光谱差异为0. 030 0、0. 031 8和0. 033 6,预测标准差为1. 933 8、0. 440 2和2. 130 9。结果表明,LAR-SLRDS算法不仅能够消除主、从仪器光谱之间存在的差异,实现模型传递,而且能够提高PLS定量模型的准确性和稳定性,具有广泛的应用潜力。
A novel model transfer method based on least angle regression(LAR) combined with simple linear regression direct standardization( SLRDS) was proposed to solve the problem of poor generality in model analysis of near infrared spectroscopy technology. In this method,the near infrared spectral data of the samples were preprocessed by wavelet transform,then the wavelength points for the spectral characteristic of the pre-processed samples were screened by LAR. Finally,the selected wavelengths of the samples were corrected by SLRDS. Near-infrared spectroscopy data for gasoline and drug samples were used to verify the performance of LAR-SLRDS. The spectral differences for the gasoline datasets C7,C8,C9 and C10 were 0. 002 8,0. 002 7,0. 002 6 and 0. 002 7,and the prediction standard deviations were 0. 410 6,0. 849 2,1. 034 9 and 1. 215 8,respectively. The spectral differences in activity,hardness and weight components of the drug datasets were 0. 030 0,0. 031 8 and 0. 033 6,and the predicted standard deviations were 1. 933 8,0. 440 2 and 2. 130 9,respectively. The experimental results showed that the LAR-SLRDS algorithm could not only eliminate the difference between the main and instrument spectra,but also improve the accuracy and stability of the PLS quantitative model.
引文
[1]Zou W B,Yin L H,Hu C Q.Chin.Pharm.(邹文博,尹利辉,胡昌勤.中国药房),2018,(3):416-420.
[2]Kong Q Q,Ding X Q,Gong H L,Li Z R,Tang X H,Yu C X.J.Instrum.Anal.(孔清清,丁香乾,宫会丽,李忠任,唐兴宏,于春霞.分析测试学报),2017,36(10):1203-1207.
[3]Zhao X,Zhang R,Wang W,Li C Y.Food Sci.,2018,39(8):249-255.
[4]Risoluti R,Materazzi S,Gregori A,Ripani L.Talanta,2016,153:407-413.
[5]Zhang X B,Feng Y C,Hu C Q.Chin.J.Pharm.Anal.(张学博,冯艳春,胡昌勤.药物分析杂志),2009,(8):1390-1399.
[6]Folch-Fortuny A,Vitale R,Denoord O E,Ferrer A.J.Chemom.,2017,31(3):e2874.
[7]Soldado A,Fearn T,Martínez-Fernández A,Roza-Delgado B.Talanta,2013,105(4):8-14.
[8]Bouveresse E,Massart D L.Vib.Spectro.,1996,11(1):3-15.
[9]Wang Y D,Veltkamp D J,Kowalski B R.Anal.Chem.,1991,63(23):530-533.
[10]Shenk J S,Westerhaus M O,Templeton W C.Cropscience,1985,25(1):159-161.
[11]Yang H,Zhang X F,Fan Y X,Xie P M,Chu X L.Chin.J.Anal.Chem.(杨辉华,张晓凤,樊永显,谢谱模,褚小立.分析化学),2014,42(9):1229-1234.
[12]Blank T B,Sum S T,Brown S D,Monfre S L.Anal.Chem.,1996,68(17):2987-2995.
[13]Efron B,Hastie T,Johnstone I,Tibshirani R.Ann.Statist.,2004,32(2):407-451.
[14]Xiong Q,Zhang R Q,Li H,Chen W C,Du Y P.J.Instrum.Anal.(熊芩,张若秋,李辉,陈万超,杜一平.分析测试学报),2018,37(7):778-783.
[15]Wang J X,Li H,Xing Z N,Guo H G.J.Instrum.Anal.(王菊香,李华,邢志娜,郭恒光.分析测试学报),2011,30(1):43-47.
[16]Ni L J,Xiao L X,Zhang L G,Luan S R.J.Instrum.Anal.(倪力军,肖丽霞,张立国,栾绍嵘.分析测试学报),2018,37(5):539-546.
[17]Chen D,Cai W,Shao X.Anal.Chim.Acta,2007,598(1):19-26.
[18]Oliveri P,Casolino M C,Casale M,Medini L,Mare F,Lanteri S.Anal.Chim.Acta,2013,761:46-52.
[19]Bin J,Li X,Fan W,Zhou J H,Wang C W.Analyst,2017,142(12):2229-2238.
[20]Wang J J,Zhe W,Liu Y,Cai W S,Shao X G.Acta Tabacaria Sinica(王家俊,者为,刘言,蔡文生,邵学广.中国烟草学报),2014,20(6):1-5.