摘要
运用拉曼光谱技术和基于角度度量的多变量分析方法,对在线监测聚氨酯预聚物聚合过程中异氰酸酯基团(—NCO)含量进行研究。通过采集反应过程中样本的—NCO含量数据及反应体系拉曼光谱数据,将拉曼强度数据转换为角度度量数据,并对该角度度量数据和—NCO含量实测值进行偏最小二乘回归分析,建立了聚氨酯预聚物合成过程—NCO基团的定量分析模型。预测结果与实际含量相比,最大绝对误差为0.20%,平均绝对误差为0.06%,预测均方根误差为0.07,相关系数达0.999,满足无损、高效、准确检测的要求,为实现在线监测—NCO含量提供了一种有效的方法。
The on-line monitoring of --NCO percentage in polyurethane prepolymer during polymerization process was studied by using Raman spectroscopy and multivariate analysis method of angle measurement. The Raman intensity data of polyurethane reaction was collected and converted to angle measurement data, then the angle measurement data and the measured --NCO percentage were regression analyzed to establish partial least-squares （PLS） model of --NCO percentage in polyurethane prepolymer. The predicted results were very satisfactory： the --NCO percentage was predicted with a maximum absolute error of prediction 0. 20%, a mean absolute error 0. 06%, a root mean squared error of prediction 0. 07%, and a correlation coefficient 0. 999. It is qualified to the lossless, efficient and accurate testing requirement, providing an effective way for the realization of on-line monitoring of --NCO percentage.
出处
《涂料工业》
CAS
CSCD
北大核心
2015年第8期57-60,66,共5页
Paint & Coatings Industry
基金
广西科技计划项目(桂科攻1355010-15)
广西自然科学基金2014GXNSFAA118056
