电子鼻对低温贮藏猕猴桃品质的预测

为了探索电子鼻技术快速检测猕猴桃品质的方法，以“秦美”猕猴桃为试材，利用电子鼻技术对低温贮藏猕猴桃的芳香成分进行检测，采用多元线性回归（multiple linear regression，MLR）、偏最小二乘法（partial leastsquaresregressions，PLS）、BP（back-propagation）网络3 种分析方法建立评价低温贮藏期猕猴桃的可溶性固形物含量、pH值和硬度的数学模型。结果表明：在贮藏0～45 d，S1、S2、S3、S4、S7、S8、S9和S10传感器响应值变化显著（P＜0.05），即芳香苯类、氮氧化物、氨类、氢气、硫化氢、乙醇、有机硫化物、芳香烷烃这几类化合物在猕猴桃低温贮藏期变化显著。同时线性判别分析比主成分分析能更好地区分不同贮藏期的猕猴桃。MLR、PLS和BP网络3 种分析方法都能很好地预测低温贮藏猕猴桃的品质，但相比之下，BP网络的分析精度更高。应用电子鼻技术预测猕猴桃的品质是可行的。

Prediction of Kiwifruit Quality during Cold Storage by Electronic Nose

In order to explore the applicability of electronic nose technique for rapid and non-destructive evaluation of
kiwifruit quality, the volatile compounds of “Qinmei” kiwifruit during cold storage were studied by electronic nose. Multiple
linear regression (MLR), partial least-squares regression (PLS) and back-propagation (BP) network were applied to predict
the firmness, soluble solid content (SSC) and pH of kiwifruit based on the signal of electronic nose. The results showed that
the response values of sensors S1, S2, S3, S4, S7, S8, S9 and S10 were relatively high and changed significantly during 45
days of storage (P ＜ 0.05). In addition, aromatic benzene, nitrogen oxide, ammonia, hydrogen, hydrogen sulfide, ethanol,
organic sulphur compounds and aromatic alkane also exhibited a significant change during cold storage. Linear discriminant
analysis was able to better distinguish among different storage periods of kiwifruit than principal component analysis.
PLS, MLR and BP network were able to predict the firmness, soluble solid content and pH of kiwifruit during cold storage.
However, BP network led to more precise predictions than PLS and MLR. The results indicate that it is possible to use this
non-destructive technique for measuring quality characteristics of kiwifruit, and electronic nose technique provides a method
for rapid and non-destructive evaluation of kiwifruit quality.