一种基于改进偏最小二乘算法的中红外光谱测定在用润滑油酸值的新方法

摘要

提出一种基于改进偏最小二乘法——多主成分数偏最小二乘算法的中红外光谱快速测定在用润滑油酸值的新方法.通过提取在用润滑油样品中红外光谱指纹区( 600～1870 cm-1)谱图数据的特征信息,建立酸值的多元校正模型,并对模型进行全交互验证和验证集样品评价,评价结果证明了该方法的可行性和有效性.%A new method based on mid -infrared spectral data and improved partial least squire(PLS)to determine the acid number( AN)of used lubricating oil,namely multiple principal components PLS,was put forward. Multivariate calibration models for AN properties were built based on the fingerprint band ranging from 600 cm-1 to 1 870 cm-1. Full cross validation and validation samples were adopted to evaluate the PLS models,whose results prove that the new method is feasible and effective.