摘要
应用近红外光谱技术(NIRS),采用偏最小二乘法(PLS)建立发酵液中乙醇含量定量分析预测模型,并比较发酵液原液和发酵液上清液NIRS预测模型对预测乙醇含量的效果差异。结果表明,发酵液原液所建立的乙醇含量和发酵上清液所建立的乙醇含量预测模型校正误差(SEC)、交叉检验标准误差(SECV)、校正相关系数(RSQ)、交叉验证相关系数(1-VR)皆有很好的关联性。由此可见,通过内部交叉检验和外部验证.用近红外光谱法测定发酵液中的乙醇含量具有很高的准确度,为近红外光谱法快速监测乙醇发酵提供了新的方法。
With near infrared reflectance spectroscope （NIRS）, models for quantitative analysis of alcohol content in different types of fermentation broth were established by using partial least square （PLS） method. The results indicated good correlations among the models by the standard errors （SEC） for calibration, SECV for cross validation and determination coefficients RSQ for calibration and 1-VR for calibration. The method of NIRS has high accuracy in predicting tests of cross validation and independent validation, thus offering a new and efficient way for quick analysis/on-line monitoring of fermentation alcohol production.
出处
《食品与发酵工业》
CAS
CSCD
北大核心
2008年第7期136-140,共5页
Food and Fermentation Industries
关键词
近红外光谱
发酵
乙醇含量
预测模型
near infrared reflectance spectroscopy , fermentation, alcohol, calibration model
