Next Article in Journal
A 100-km BOFDA Assisted by First-Order Bi-Directional Raman Amplification
Previous Article in Journal
Biosensors for Epilepsy Management: State-of-Art and Future Aspects
Previous Article in Special Issue
Freshness Evaluation of Three Kinds of Meats Based on the Electronic Nose
Open AccessArticle

Electronic Nose-Based Technique for Rapid Detection and Recognition of Moldy Apples

by Wenshen Jia 1,2,3,4, Gang Liang 1,2,3,*, Hui Tian 1,2,3,4, Jing Sun 5,6 and Cihui Wan 1
1
Beijing Research Center for Agricultural Standards and Testing, Beijing Academy of Agriculture and Forestry Science, Beijing 100097, China
2
Risk Assessment Lab for Agro-products (Beijing), Ministry of Agriculture, Beijing 100097, China
3
Beijing Municipal Key Laboratory of Agriculture Environment Monitoring, Beijing 100197, China
4
Key Laboratory of Urban Agriculture (North China), Ministry of Agriculture and Rural Affairs, Beijing 100097, China
5
Chinese Academy of Agricultural Engineering, Beijing 100121, China
6
Key Laboratory of Agro-Products Postharvest Handling, Ministry of Agriculture, Beijing 100121, China
*
Author to whom correspondence should be addressed.
Sensors 2019, 19(7), 1526; https://doi.org/10.3390/s19071526
Received: 23 February 2019 / Revised: 25 March 2019 / Accepted: 25 March 2019 / Published: 29 March 2019
(This article belongs to the Special Issue Electronic Noses and Their Application)
In this study, the PEN3 electronic nose was used to detect and recognize fresh and moldy apples inoculated with Penicillium expansum and Aspergillus niger, taking Golden Delicious apples as the model subject. Firstly, the apples were divided into two groups: individual apple inoculated only with/without different molds (Group A) and mixed apples of inoculated apples with fresh apples (Group B). Then, the characteristic gas sensors of the PEN3 electronic nose that were most closely correlated with the flavor information of the moldy apples were optimized and determined to simplify the analysis process and improve the accuracy of the results. Four pattern recognition methods, including linear discriminant analysis (LDA), backpropagation neural network (BPNN), support vector machines (SVM), and radial basis function neural network (RBFNN), were applied to analyze the data obtained from the characteristic sensors, aiming at establishing the prediction model of the flavor information and fresh/moldy apples. The results showed that only the gas sensors of W1S, W2S, W5S, W1W, and W2W in the PEN3 electronic nose exhibited a strong signal response to the flavor information, indicating most were closely correlated with the characteristic flavor of apples and thus the data obtained from these characteristic sensors were used for modeling. The results of the four pattern recognition methods showed that BPNN had the best prediction performance for the training and testing sets for both Groups A and B, with prediction accuracies of 96.3% and 90.0% (Group A), 77.7% and 72.0% (Group B), respectively. Therefore, we demonstrate that the PEN3 electronic nose not only effectively detects and recognizes fresh and moldy apples, but also can distinguish apples inoculated with different molds. View Full-Text
Keywords: electronic nose; apple; mildew; pattern recognition; artificial neural network; nondestructive examination electronic nose; apple; mildew; pattern recognition; artificial neural network; nondestructive examination
Show Figures

Figure 1

MDPI and ACS Style

Jia, W.; Liang, G.; Tian, H.; Sun, J.; Wan, C. Electronic Nose-Based Technique for Rapid Detection and Recognition of Moldy Apples. Sensors 2019, 19, 1526.

Show more citation formats Show less citations formats
Note that from the first issue of 2016, MDPI journals use article numbers instead of page numbers. See further details here.

Article Access Map by Country/Region

1
Back to TopTop