摘要
采用中红外光谱分析技术结合化学计量学方法对酱香型白酒轮次酒中的总酸及总酯含量进行定量分析,采用标准正态变量(SNV)对光谱进行预处理,应用遗传算法(GA)分别结合偏最小二乘法(PLS)及最小二乘支持向量机(LS-SVM)建立定标模型,并采用独立测试集样品对模型准确度进行验证。结果表明,采用标准正态变量(SNV)对样品光谱进行预处理可提高建模效果,而经遗传算法(GA)对特征变量优化后建立的偏最小二乘(PLS)模型分别将总酸、总酯两项指标模型的变量数降低至38及25个。以GA优选的波长作为输入变量建立的最小二乘支持向量机(LS-SVM)模型效果更优,总酸及总酯模型的决定系数(R2)分别提升至0.994及0.987,预测均方根误差(RMSEP)分别降低至0.031、0.082,相对分析误差(RPD)值分别为11.94、8.05。验证实验结果表明,总酸及总酯模型的决定系数(R2)分别为0.993、0.985,RMSEP分别为0.036、0.089,表明建立的定标模型准确度及稳定性良好,可快速检测酱香型白酒中的总酸和总酯含量,为液体白酒品质的快速分析方法研究提供技术借鉴。
The content of total acid and total ester in sauce-flavor Baijiu round liquor was quantitatively analyzed by mid-infrared spectroscopy combined with chemometrics method.Standard normal variable(SNV)was used to pretreat the spectrum,and genetic algorithm combined with partial least squares(PLS)and least squares support vector machine(LS-SVM)was used to establish the calibration model,and independent test set samples were used to verify the accuracy of the model.The results showed that the SNV could improve the modeling effect,and the PLS model established by genetic algorithm after optimizing the characteristic variables reduced the number of variables of total acid and total ester to 38 and 25 respectively.The LS-SVM model with genetic algorithm preferred wavelength as input variable had better results,the coefficient of determination(R2)of total acid and total ester models was increased to 0.994 and 0.987,respectively,and the root mean square error(RMSEP)of prediction was reduced to 0.031 and 0.082,respectively.The relative percent deviation(RPD)values were 11.94 and 8.05,respectively.Meanwhile,the RPD values of total acid and ester reached 11.94 and 8.05,indicating that the established calibration model had good accuracy and stability,it can quickly detect the content of total acid and total ester in sauce-flavor Baijiu,which provided technical reference for the rapid analysis method of Baijiu quality.
作者
唐云容
庞臻
钟敏
路虎
潘成康
文永
李子文
TANG Yunrong;PANG Zhen;ZHONG Min;LU Hu;PAN Chengkang;WEN Yong;LI Ziwei(Guizhou Xijiu Co.,Ltd.,Zunyi 564622,China;China Food Fermentation Industry Research Institute Co.,Ltd.,Beijing 100015,China)
出处
《中国酿造》
CAS
北大核心
2024年第1期244-249,共6页
China Brewing
基金
贵州省科技支撑计划(黔科合支撑【2023】一般342)。
关键词
酱香型白酒
中红外光谱分析技术
化学计量学方法
定量分析
sauce-flavor Baijiu
mid-infrared spectroscopy analysis
chemometrics method
quantitative analysis
