基于BP神经网络的电子鼻羊奶贮藏时间的预测

利用电子鼻PEN3系统判定室温和冷藏条件下羊奶的贮藏时间。通过电子鼻系统采集羊奶室温贮藏及冷藏期间挥发性成分的响应值,并采用PCA(主成分分析法)、LDA(线性判别分析法)和LM算法优化的BP神经网络(LM-BP)、遗传算法优化的神经网络(GANN)、4层BP神经网络进行模式识别。结果表明PCA和LDA均可区分室温贮藏及冷藏1～6d的生鲜羊奶,LDA方法还可以明显体现出羊奶贮藏期间挥发性成分的变化趋势,并且与羊奶酸度的变化有很好的一致性。采用LM-BP神经网络、GANN神经网络和4层神经网络均能较好地预测不同贮藏时间的羊奶,其中4层神经网络的预测正确率高于LM-BP神经网络和GANN神经网络。

Discrimination of fresh goat milk during storing based on electronic nose by BP neural network

The classification of different storage time of fresh goat milk at ambient temperature and refrigerated temperature was detected by an electronic nose. The volatile composition emanating from the goat milk was sampled by PEN3 systems,and the response values of PEN3 were obtained. The data was analyzed using PCA(principal component analysis)and LDA(linear discrimination analysis),the combination of a Levenberg - Marquardt algorithm and BP neural network(LM - BP),the combination of a genetic algorithm and BP neural network(GANN)and 4 layer BP neural network. The results showed that electronic nose was able to classify fresh goat milk during 1 ~ 6d storing at ambient temperature and refrigerated temperature by PCA and LDA,however LDA showed variation trend of volatile composition clearly which had a great agreement with acidity of the samples. Better prediction values were obtained by 4 layer BP neural network than GANN and LM - BP.