机构地区: 广西科技大学生物与化学工程学院
出 处: 《化学分析计量》 2012年第6期32-35,共4页
摘 要: 阿司匹林合成过程中所采集的阿司匹林、水杨酸和乙酸酐的拉曼光谱存在交迭现象,很难从光谱数据中分离各组分含量。根据反应体系的组成和浓度配制样本溶液并采集拉曼光谱,运用多种方法对光谱进行预处理结合偏最小二乘法(PLS)对混合体系中阿司匹林、乙酸酐和水杨酸3组分浓度和光谱矩阵建立回归模型,对回归结果进行比较,根据交互验证均方根误差RMSECV,预测相关系数R2,预测均方根误差RMSEP。结果表明预处理方案可以有效降低基线漂移对预测精度的影响,根据Q值检验法和T2检验法确定经矢量归一+Savitzky-Go1ay五点二次平滑一阶卷积求导+均值中心化对拉曼光谱预处理,PLS主成分选4,交互验证选用"逐一法"可对阿司匹林反应体系3组分进行同时测定,此训练集为阿司匹林合成体系的拉曼光谱在线监测提供了良好的数据支持。 The existence of overlapping phenomena in the raman absorption of aspirin, sallcyuc acltl ana acenc anhydride made it difficult to use raman spectroscopy to determine the component concentration of aspirin, salicylic acid and acetic anhydride in synthesis process of aspirin. According to the amount of material changes before and after reaction made concentration sample build correction set and test set, and it' s raman information was extracted and analyzed by applying a variety of spectral pretreated methods combined with partial least squares establish and salicylic acid, acetic anhydride determination and acetyl salicylic acid quantitative analysis model and realized synthetic acetylsalicylic acid system of components concentration prediction. The result of the root mean squares error of cross-validation (RMSECV), the correlation coefficient of test set and the root mean squares error of prediction set (RMSEP) indicated that the influence of the prediction precision by the baseline drift could be effectively reduced. According to Q Residual and Hotelling T2 the method of simultaneous determination of three components in synthesis systems of aspirin by raman spectroscopy which PLS principal factor was 4, and number of Data Splits of contiguous block by Cross-Val method was "leave one out" after raman spectra pretreatment by the normalize+Savitzky-Go lay five points two times smoothing first-order derivation+mean center was determined. This method provides effective data support for on-line raman spectroscopy monitoring of aspirin synthesis.