摘要
使用傅里叶变换近红外光谱(FT-NIR)结合化学计量学方法,开发了一种黑果枸杞干果和鲜果中主要成分(总糖、还原糖、总酸、氨态氮、花青素、原花青素、总酚、黄酮和多糖)含量的预测方法.首先比较了 11种原始光谱的预处理方式,筛选出每种成分的最优预处理方法.然后比较了利用偏最小二乘(PLS)、区间偏最小二乘(iPLS)和联合区间偏最小二乘(siPLS)算法建立的模型,最终确定采用siPLS建模.结果表明:总酸、氨态氮、花青素、原花青素、总酚和黄酮的交叉验证相关系数(Rc)和预测集相关系数(RP)均大于 0.9818,相对分析误差(RPD)均大于 2.5,模型效果优异,总糖、还原糖和多糖的建模效果良好,建立的定标模型均可以用于实际检测.验证集样本实测值与预测值无显著性差异,预测误差在±0.1%,模型的预测结果可信度高.本研究建立的预测模型,可以实现黑果枸杞干果和鲜果中主要成分含量的无损、快速、准确检测.
Abstract
A quantitative method of main components(total sugar,reducing sugar,total acid,ammonia nitrogen,anthocyanins,procyanidins,total phenols,flavonoids and polysaccharides)in Lycium ruthenicum Murr was developed using Fourier transform near infrared spectroscopy(FT-NIR)combined with chemometric analysis.Firstly,11 pretreatment methods were compared for the original spectra,and the optimal pretreatment method of each component was selected.Then,the model results established by partial least squares(PLS),interval partial least squares(iPLS)and synergistic interval partial least squares(siPLS)algorithms were compared,and finally siPLS was adopted for modeling.The results showed that the correlation coefficient of calibration(Rc)and prediction(RP)of total acids,ammonia nitrogen,anthocyanins,procyanidins,total phenols and flavonoids were all greater than 0.9818,and the relative analysis error(RPD)was more than 2.5,indicating an excellent model performance.The modeling effect of total sugar,reducing sugar,and polysaccharide was also good.The established calibration models could be used for actual detection.The predicted values of the verification samples did not significantly differ from the measured values,with a prediction error of only±0.1%.Therefore,the prediction model had high reliability.The prediction model established in this study can enable nondestructive,rapid and accurate main components in dried and fresh fruits of Lycium ruthenicum Murr.
基金项目
国家重点研发计划项目(2021YFD2100102-4)
国家自然科学基金重点项目(22138004)
国家自然科学基金青年项目(32001828)
青海省重大科技专项(2020-SF-A2)
维普
万方数据