一种面向PX装置成品杂质含量检测的拉曼光谱分析法

A Detection Method for Impurity Content of PX Product Based on Raman Spectrum

面向对二甲苯(PX)成品的在线分析问题,为克服色谱分析法的不足,本文提出了一种基于拉曼光谱的PX成品杂质含量分析方法。为了克服激光中心波长与功率波动带来的不良影响,提出了激光中心波长自动校正与PX特征峰归一化方法;同时,结合相关分析获取了主要杂质的特征波段,并采用偏最小二乘算法进行定量分析。针对11个未知样本,PLS模型对甲苯(MB)、乙苯(EB)、间二甲苯(MX)、邻二甲苯(OX)的标准预测误差分别为0.03%、0.11%、0.02%、0.03%。实验结果表明,拉曼光谱法能够快速、准确且无损地测定PX成品中主要杂质的含量,为在线拉曼分析系统的开发奠定了方法基础。

This paper proposed an analysis method to calculate the impurity content in p- xylene（PX） based on Raman spectroscopy. To overcome the disturbance caused by the fluctuation of laser wavelength and power, laser wavelength automatic calibration and PX characteristic peak normalized method are applied. Feature band of the impurity content was located by correlation analysis. The partial least squares （PLS） regression algorithm was used to build the quantitative analysis model within the feature band. The detection errors of toluene, ethylbenzene, m-xylene and o-xylene are ±0.03 %, ± 0. 11%, ± 0.02 % and ±0.03%. The experimental result shows that this method can be an non-destructive method to determine the impurity content of PX product quickly and accurately.