Premium
Classification of impact injury of apples using electronic nose coupled with multivariate statistical analyses
Author(s) -
Ren Yamei,
Ramaswamy Hosahalli S.,
Li Ying,
Yuan Chunlong,
Ren Xiaolin
Publication year - 2018
Publication title -
journal of food process engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.507
H-Index - 45
eISSN - 1745-4530
pISSN - 0145-8876
DOI - 10.1111/jfpe.12698
Subject(s) - electronic nose , principal component analysis , linear discriminant analysis , artificial intelligence , pattern recognition (psychology) , multivariate statistics , artificial neural network , statistics , mathematics , discriminant function analysis , computer science
An electronic nose equipped with a headspace sampling unit was evaluated as a non‐destructive method for determining damage degree. Fuji apples were dropped from different heights (0.2–0.8 m) onto a cement floor inflict damages. E‐nose data was evaluated by Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) to distinguish apples based on damage severity. LDA performed better than PCA for classifying the apples. Stepwise Discriminant Analysis (SDA), Radial Basis Function Neural Network (RBFN), Multilayer Perceptron Neural Networks (MLPN), and Back‐Propagation Neural Network (BPNN) models were employed for pattern recognition. With SDA dataset, the correct classification rate (CCR) was 97.5% for training and 93.8% testing; MLPN resulted in 100%, and RFBN performed better only with more severe damages. The BPNN model had excellent correlation with classification values for damaged apples ( R 2 > 0.98). Therefore, E‐nose technology with ANN and multivariate statistics is an effective way for classifying damaged apples.