A modified mean deviation threshold function based on fast Fourier transform and its application in litchi rest storage life recognition using an electronic nose

Since gas sensor drift is a main limitation for the application of an electronic nose, and a reference standard is necessary for shelf management of litchi fruit, a modified mean deviation threshold function based on fast Fourier transform (MDFF-FFT) for electronic nose drift elimination and a new concept the rest storage life (RSL) for litchi fruit shelf situation evaluation have been constructed in this study. Three commonly used threshold acquisition methods, unbiased estimator, fixed threshold, and mini-max principle were evaluated to instead of selecting threshold value randomly for present MDFF-FFT. A PEN3 portable electronic nose was applied to recognize the RSL of litchi during storage across room temperature (RT), refrigerator environment (RE) and controlled-atmosphere (CA) environments. Linear discriminant analysis (LDA), probabilistic neural network (PNN), and partial least squares regression (PLSR) were used to compare the RSL classification effect, recognition accuracy, and predict ability of litchi stored in the three environments based on electronic nose with the drift elimination of different threshold acquisition methods using MDTF-FFT. The results showed that an electronic nose has the potential to recognize the RSL of litchi stored in different environments. Unbiased estimator method can provide better threshold than other threshold acquisition methods for MDTF-FFT. After drift elimination by unbiased estimator method combined with MDTF-FFT, litchi RSL can be classified, recognized and predicted by electronic nose effectively, the accuracy of which was higher than control (no drift elimination) and drift elimination with other methods.