基于偏最小二乘法的板栗近红外光谱分析模型的建立

Establishment of NIR models for components determination in fresh chestnut

目的:应用近红外光谱技术和化学计量学方法,建立板栗品质分析的近红外光谱模型。方法:采用傅里叶变换近红外光谱仪,采集样品的近红外漫反射光谱,再用传统理化分析方法测得样品的各项品质参数,采用偏最小二乘法(PLS)建立定标模型,内部交叉验证法对模型进行检验。结果:对板栗分别建立了水分、淀粉、硬度和糖度的PLS模型,4种PLS模型都非常理想,模型的相关系数均大于0.99。结论:采用近红外光谱法可以实现板栗品质指标的快速无损检测。

Objective：To quantitative determinate several components of fresh chestnut directly and rapidly by using near infrared spectroscopy and chemometrics.Method：The NIR spectra of calibration set samples were collected with FT-NIR spectrometer,and the reference concentrations of components were analyzed with the standard chemical analysis methods.The calibration models were established using partial least square（PLS） regression and validated with leave-one-out cross validation method.Result：Result：The calibration models for analysis of water,amylum,texture and sugar contents were established with correlation coefficients over 0.99.Conclusion：It is possible to use NIR method to rapidly determinate the content of the above components in fresh chestnut.