| 摘要: |
| 【目的】研究利用电子鼻对苹果低温贮藏时间及品质的预测效果，为苹果低温贮藏品质的无损检测及合理加工利用提供参考。【方法】以富士苹果为试材，在（0±1） ℃低温条件下贮藏，分别在贮藏后的第0，30，60，90，120，150和180天，随机选取30个果实，利用PEN3型电子鼻检测其香气，并一一对应测定苹果的主要品质指标（硬度、可溶性固形物含量和可滴定酸含量）。利用载荷分析优化电子鼻传感器阵列，对优化后的电子鼻检测数据进行线性判别分析，建立苹果低温贮藏品质的偏最小二乘预测模型、BP神经网络预测模型和贮藏时间的多层感知器预测模型，并对预测效果进行了比较。【结果】线性判别分析能够较好地区分苹果的贮藏品质，且苹果香气在贮藏60~90 d时变化较大；建立的多层感知器神经网络模型对苹果贮藏时间有较好的预测效果，预测准确率均＞92.0%；利用偏最小二乘法和BP神经网络均能对果实的品质建立有效的预测模型，其中偏最小二乘法对冷藏苹果硬度和可滴定酸含量的预测效果优于对可溶性固形物含量的预测，利用BP神经网络所建立预测模型的决定系数均＞0.930 0，预测效果较偏最小二乘法更好。【结论】利用电子鼻的快速无损检测功能可以实现对苹果低温贮藏时间及品质的预测。 |
| 关键词: 电子鼻 苹果 低温贮藏 贮藏时间 品质预测 |
| DOI： |
| 分类号: |
| 基金项目:国家现代苹果产业技术体系专项（NYCYTX-08-05-02） |
|
| Prediction of low-temperature storage time and quality of apples based on electronic nose |
|
LI Ying,REN Ya-mei,ZHANG Shuang,et al
|
| Abstract: |
| 【Objective】The effect of electronic nose (E-nose) on prediction of storage time and quality of apples at low-temperature ((0±1) ℃) was studied to provide the reference for nondestructive test of apple quality and reasonable processing and utilization of apples.【Method】 “Fuji” apples were stored in low-temperature (0±1) ℃.30 apples were picked randomly to determine their aroma by E nose of PEN3 and quality indices (firmness,soluble solid content and titratable acidity) by traditional methods at 0,30,60,90,120,150,and 180 days after storage.Loading analysis was used to optimize electronic nose sensors array,and the optimized data was analyzed by linear discriminant analysis before partial least squares and BP neural network models were established for quality prediction and multilayer perceptron model was established for storage time prediction.At last,the results were compared with observation.【Result】Storage quality of apples can be accurately predicted by linear discriminant analysis (LDA) and the aroma of apples changed significantly at 60-90 d.Storage times were accurately obtained by multilayer perceptron neural network (MLPN) with the accuracy of >92.0%.Good relationships were established between the E-nose signals and apple quality by partial least squares regression (PLS) and BP neural network (BPNN).PLS had better performance at predicting firmness and titratable acidity than predicting soluble solid content.BPNN had better performance than PLS with determination coefficient of >0.930 0.【Conclusion】Electronic noses can be used as a rapid and nondestructive testing technology to predict storage time and quality of apple stored at low-temperature. |
| Key words: electronic nose apple low-temperature storage storage time quality prediction |
