近红外光谱的润滑油黏度波段筛选研究

Research on the Near-infrared Spectroscopy Wave Band Method for Screening Viscosity of Lubricating Oil

针对传统方法测量润滑油黏度时操作要求高、不易推广等问题,提出一种基于近红外光谱的润滑油黏度波段筛选方法。选取不同黏度值润滑油为样本,采用偏最小二乘(PLS)定量建模方法,对比分析不同光谱预处理方法,去除与目标因素无关的信息。结合间隔偏最小二乘(iPLS)和遗传算法(GA)对建模变量进行筛选,将筛选的变量作为PLS模型输入变量。结果表明,预测集系数与均方根误差均达到建模精度,为实现快速判别提供了参考价值。

In traditional viscosity measurement of lubricating oil, the operational requirements are rigorous, so the method is not easy to promote. Against this problem, the method based on near-infrared spectroscopy wave band for screening viscosity is proposed. With the lubricating oil in different viscosity as samples, by means of partial least square （ PLS） quantitative modeling method, comparative analysis for different spectroscopy pretreatment methods is conducted, to eliminate the information that is not related to the objective factors. Combining the interval partial least square （ IPLS） and genetic algorithm （ GA） , the modeling variables are filtered, and the screened variables are used as the input variables of the PLS model. The results show that both the prediction set coefficient and root mean square error of cross validation （ RMSECV） meet the modeling accuracy, thus the reference is provided for fast speed judgment.