摘要
采用近红外光谱分析可满足糖蜜发酵酒精生产中酒精含量实时检测需求。本文以蒸出酒精后的发酵原液作参比获取的近红外透射光谱,通过最小二乘法建立酒精含量定量模型,用"残差平方和"和"独立变量数"两种方法判断模型主成分数,模型预测能力评价表明"独立变量数"判断更准确。所建模型具有良好的精确度和稳定性,可用于糖蜜酒精发酵的常规检测,符合酒精发酵过程的在线监测要求。
Near Infrared Spectroscopy Analysis meets the demand of real-time determination of alcohol in molasses fermentation broth. The near-infrared transmission spectrum were fetched from fermentation broth and the references were selected as the fermentation broth which had been steamed out of alcohol.The quantitative models of alcohol was molded using the Partial Least Squares(PLS) method.The numbers of Principal Components(PCs) for PLS were discussed following " Prediction Residual Error Sum of Square, PRESS" and " Number of Independent Variables, NIV".The model predictive capability showed that the number of PCs from NIV was more accurate.The model had good accuracy and stability and could be used for normal determination of molasses alcoholic fermentation, but it also fit the on-line monitoring for the alcoholic fermentation process.
出处
《食品工业科技》
CAS
CSCD
北大核心
2010年第11期369-371,共3页
Science and Technology of Food Industry
基金
广西科学基金(0832064)
广西教育厅科研项目(200608MS055
200708MS067)
关键词
糖蜜发酵液
近红外光谱
酒精
偏最小二乘法
独立变量数
molasses fermentation broth
near-infrared spectroscopy
alcohol
partial least squares
number of independent variable