关键词: 苹果, 水心病, 近红外光谱, 电子鼻, 化学计量学

Abstract: This study aimed to explore the feasibility of applying near infrared spectroscopy (NIR) and electronic nose (E-nose) for detecting apple watercore. A total of 277 samples of “Qinguan” apples with watercore and healthy apples were tested. NIR spectra of each sample in the range of 12 000 to 4 000 cm-1 and E-nose signals from 10 sensors were collected, The Fisher discriminant model was established with the principal components extracted by different preprocessing methods. Meanwhile, the E-nose data were used for modeling by 3 different chemometric methods. The results indicated that the Fisher discriminant model developed based on the first twenty principal components from the NIR spectra subjected to the first derivative (9-point smoothing) pretreatment worked best with discrimination accuracy rates of 100% for unknown samples. The correct discrimination rates of the discriminant models developed by Fisher discriminant, multilayer perceptron (MLP) neural network and radial basis function (RBF) neural network for unknown samples were 89.7%, 89.5% and 85.7%, respectively. Thus, the combined application of NIR spectroscopy and E-nose with chemometrics can rapidly and nondestructively test watercore apples. NIR spectroscopy combined with Fisher discriminant analysis is an accurate and reliable method for detecting watercore apples with the highest correct recognition rate.

Key words: apple, watercore, NIR spectroscopy, electronic nose, chemometrics