Fresh peach juice and commercial juices were assessed by a trained sensory panel. Differences in the volatiles between fresh and commercial juices were distinguished by partial least squares regression (PLSR) combined with backpropagation artificial neural network. Totally, 10 compounds of three samples were identical: five esters, four alcohols, and one aldehyde. In addition, principal component analysis and hierarchical clustering analysis combined with PLSR revealed that esters were the characteristic volatiles accounting for the differences between fresh and commercial juices. By taking esters as input, the established BP network could correctly distinguish flat peach and commercial juices. Moreover, 10 esters were selected as the key volatiles with the BP network. However, commercial treatment caused a significant reduction in the content of the key esters between fresh flat peaches and commercial juices. Therefore, the changes in esters caused the significant differences in flavor quality between fresh juice and flat peach juices products.

中文翻译：

---

PLSR和BP网络对商业加工对扁桃汁中挥发性化合物和感官特征的影响

新鲜的桃汁和市售果汁由训练有素的感官小组评估。新鲜果汁和商用果汁之间的挥发物差异通过偏最小二乘回归（PLSR）结合反向传播人工神经网络来区分。总共，三个样品中的10种化合物是相同的：五种酯，四种醇和一种醛。此外，主成分分析和层次聚类分析与PLSR结合发现酯是特征性的挥发物，解释了新鲜果汁和商用果汁之间的差异。通过将酯类作为输入，已建立的BP网络可以正确地区分桃子汁和市售果汁。此外，在BP网络中，选择了10种酯作为关键挥发物。然而，商业化处理导致新鲜扁桃和商业果汁之间的关键酯含量大大降低。因此，酯类的变化导致鲜榨果汁和扁桃果汁产品之间的风味品质出现显着差异。