NIR光谱结合LLE-PLS建模用于安神补脑液提取过程分析的研究

Application of near infrared spectra coupled with LLE-PLS modeling to extraction process of Anshen Bunao Syrup

目的:探讨安神补脑液提取过程近红外(NIR)光谱分析的可行性及其方法。方法:提出NIR光谱建模的局部线性嵌入(LLE)-偏最小二乘(PLS)方法,该方法首先用LLE对NIR光谱数据降维,再用PLS建立校正模型。结果:参数优化后的LLE-PLS,二苯乙烯苷留-法交叉验证均方根误差(RMSECV)为0.0457mg/mL、决定系数(R2)为0.9673,淫羊藿苷RMSECV为0.0333mg/mL、R2为0.9809,均优于常规的PLS建模方法。结论:基于NIR光谱和LLE-PLS方法可实现安神补脑液提取过程的在线分析。

AIM ： To discuss the feasibility and methods of the appling near infrared （NIR） analysis to the extraction process of Anshen Bunao Syrup. METHODS： A novel nonlinear modeling method for NIR spectra was proposed in the paper. Firstly, locally linear embedding （LLE） was adopted to lower the dimensions of NIR spectra, and then partial least squares （PLS） was used to build correction modeling. RESULTS ： After parameter optimization, LLE-PLS could accurately predict the concentration of chrysophenine with a minimum RMSECV of 0. 045 7 mg/mL, R2 of 0.967 3, and icariin with RMSECV of 0. 033 3 mg/mL, R2 of 0. 980 9, suggesting that LLE-PLS was more specific than the common PLS methods. CONCLUSION： Online quality control in the process of Anshen Bunao Syrup is realizable by combining NIR specrta and LLE-PLS modeling methods.