可见光谱结合偏最小二乘法对混合染料浓度的同时测定

摘要

A calibration model based on visible spectroscopy determination of component concentration in ternary mixed dyes was set up by using the partial least square regression method. The information was extracted and analyzed by applying a variety of spectral pretreatment methods. The effects of spectral pretreatment methods and the number of principal components on the predicted ability of PLS models were discussed. The optimum model was selected depending on the root mean squares error of cross-validation (RMSECV), the root mean squares error of calibration set (RMSEC), the correlation coefficient and the root mean squares error of prediction set (RMSEP). The results demonstrate that the optimum model possesses good stability and high precision. It is a rapid accurate approach for simultaneous determining component concentration in mixed dyes. It has important guiding significance for continuous online monitoring hybrid dye concentration.%采用偏最小二乘法建立测定混合染料浓度的可见光谱定量分析模型,应用多种光谱预处理方法对光谱进行信息提取和分析,分别建立定量分析模型；讨论了光谱的预处理方法和主成分数对PLS模型定量预测能力的影响；比较各个模型的内部交互验证均方根误差(RMSECV),校正均方根残差(RMSEC),校正相关系数R,预测均方根误差(RMSEP)选取最优模型.结果表明:所建立的校正模型稳健性好和预测精度高,为混合染料浓度的快速、准确和同时测定提供了新的途径,对于连续在线监测混合染料浓度具有指导意义.