摘要
采用电子舌对香醋发酵过程中总酸、不挥发酸、还原糖、氨基酸态氮进行定量分析。对比了偏最小二乘法(PLS)和人工神经网络(ANN)的不同算法,结果显示基于非线性映射的人工神经网络算法具有较好的定量精度,预测值和实测值的相关系数分别为0.8439、0.9382、0.8322和0.8558。预测标准偏差(RMSEP)分别为0.8240、0.0963、0.1482和0.5557。研究表明:电子舌能对香醋发酵产物进行定量预测,并对食醋发酵过程的监控有良好的应用前景。
The content of total acid, non-volatile acid, reducing sugar and amino acid nitrogen were determined by the electronic tongue technology. Two different methods, including partial least square （PLS） and artificial neural network （ANN） were compared. The results showed that ANN was better than PLS in quantitative precision. The correlation coefficients （R^2） of total acid, non-volatile acid, reducing sugar and amino acid nitrogen between observed and predicted values based on calibrations of ANN were 0.8439, 0.9382, 0.8322 and 0.8558, respectively. Root mean standard errors of prediction （RMSEP） were 0.8240, 0.0963, 0.1482 and 0.5557, respectively. Our results indicated that the electronic tongue could be used for quantitative analysis in vinegar production.
出处
《中国酿造》
CAS
北大核心
2009年第10期82-85,共4页
China Brewing
基金
国家自然基金资助项目(30671199)
关键词
电子舌技术
食醋
偏最小二乘
人工神经网络
electronic tongue
vinegar
partial least squares
artificial neural network
