Working Paper Article Version 2 This version is not peer-reviewed

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

Version 1 : Received: 28 February 2019 / Approved: 1 March 2019 / Online: 1 March 2019 (12:21:20 CET)
Version 2 : Received: 21 March 2019 / Approved: 21 March 2019 / Online: 21 March 2019 (09:54:00 CET)

A peer-reviewed article of this Preprint also exists.

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. 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.

Journal reference: Sensors 2019, 19, 1526
DOI: 10.3390/s19071526

Abstract

In this paper, PEN3 electronic nose was used to detect and recognize fresh and moldy apples (inoculated with Penicillium expansum and Aspergillusniger) taken Golden Delicious apples as model subject. Firstly, the apples were divided into two groups: apples only inoculated with 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, which can simplify the analysis process and improve the accuracy of 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 then applied to analyze the data obtained from the characteristic sensors, respectively, aiming at establishing the prediction model of 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 strong signal response to the flavor information, indicating were most closely correlated with the characteristic flavor of apples and thus the data obtained from these characteristic sensors was used for modeling. The results of the four pattern recognition methods showed that BPNN presented the best prediction performance for the training and validation sets for both the Group A and Group B, with prediction accuracies of 96.29% and 90.00% (Group A), 77.70% and 72.00% (Group B), respectively. Therefore, it first demonstrated that PEN3 electronic nose can not only effectively detect and recognize the fresh and moldy apples, but also can distinguish apples inoculated with different molds.

Subject Areas

Electronic nose; apple; mildew; pattern recognition; artificial neural network; nondestructive examination

Comments (0)

We encourage comments and feedback from a broad range of readers. See criteria for comments and our diversity statement.

Leave a public comment
Send a private comment to the author(s)
Views 0
Downloads 0
Comments 0
Metrics 0


×
Alerts
Notify me about updates to this article or when a peer-reviewed version is published.