利用德国PEN3电子鼻系统快速检测四种食醋陈化期。通过电子鼻采集食醋挥发性成分的响应值，利用主成分分析（PCA）、线性判别分析（LDA）、Fisher线性判别分析（FDA）和多层感知器神经网络（MLPNN）分析进行模式识别，结果表明：LDA分析效果优于PCA分析；并且随着陈化时间的延长，食醋的气味成分变化有增快的趋势，这种气味的变化规律与酸度的变化规律相符。用Fisher线性判别和多层感知器神经网络建立食醋陈化时间的预测模型，发现Fisher线性判别对凤翔醋、陇县醋、金台醋和渭滨醋陈化期的正确检测率分别为100%、100%、98%和100%；多层感知器神经网络对凤翔醋、陇县醋、金台醋和渭滨醋陈化期的正确检测率分别为100%、100%、96.92%和100%。由于正确检测率的高低得出电子鼻结合Fisher线性判别对食醋陈化期的监测结果优于多层感知器神经网络。

Four kinds of vinegars were detected by PEN3 electronic nose. The volatile compositions emanating from the vinegars were collected by the systems, and their response values were obtained. Principal component analysis (PCA), linear discrimination analysis (LDA), fisher linear discrimination analysis (FDA) and multilayer perceptron neural network (MLPNN) were used to distinguish the vinegars from different aging time. The results showed that LDA was able to identify different aging time of vinegars and the contribution rate was above 90%. The identification of vinegars from different aging time by LDA was better than that by PCA. Moreover, LDA showed that the changes of vinegars volatile composition had an increasing tendency during the aging time, and had a great agreement with the total acid of the samples. FDA and MLPNN were also employed to predict the aging time of the samples, which indicated that the FDA prediction rates of Feng Xiang, Long Xian, Jin Tai, Wei Bin vinegar were 100%, 100%, 98% and 100% respectively, while the MLPN prediction rates of these samples were 100%, 100%, 96.92% and 100%, respectively. Therefore, the prediction rate monitored by electronic nose combined with FDA was better than that with MLPNN.

中央高校基本科研业务费专项资金资助(QN2009072)