偏最小二乘方法在不同食醋判别分析中的应用

目的:建立一种由氢核磁共振技术（1H-NMR）和偏最小二乘回归分析（PLS）结合的方法来对不同种类食醋进行判别分析。方法:由1H-NMR技术测定的样本化学信息,经MestRe Nova软件转化成数字信息,将数据导入SIMCAP13.0软件中采用PLS进行建模和预测,统计其判别的准确率。结果:山西陈醋、镇江香醋和白醋在PLS得分图中能明显分开,并利用内部和外部的数据模型进行了验证,得出的预测准确率较高,所获得的RMSEE和RMSEP值也说明PLS模型对判别不同种类食醋具有较强的可预测性。结论:本分析方法在不同种类食醋的判别中具有较高的准确性,可为不同食醋的判别提供一种简单方法。 

Application of partial least squares in vinegar discriminant analysis

Objective: To establish a new method with a combination of1H- NMR technique and multivariate analysis applied for the discrimination of different vinegars.Methods: The chemical information of the vinegar was obtained using1H- NMR spectroscopy, and the spectra were transformed to a matrix using MestRe Nova software.Then the data were analyzed by PLS method using SIMCA- P ( 13.0) software. Results: The pattern recognition analysis by PLS ( Partial Least Squares) showed clear separation of different vinegar, and the model was confirmed by internal and external validations. The obtained RMSEE and RMSEP values also suggested the PLS model had strong predictability for discriminating different vinegars.Conclusion: a combination between non- targeted1H- NMR data and chemometric information obtained by the PLS regression approach was proposed to provide a highly reliable and accurate prediction model for distinguishing different vinegars.