Open Access Open Access  Restricted Access Subscription or Fee Access

Predicting the Shelf-life of Hurood Cheese Using Near Infrared Spectroscopy Based on RBF Neural Networks



Hurood, a kind of Chinese traditional cheese with high moisture, needs cold storage in the process of transportation and preservation otherwise the frequency of deterioration is very high. In order to predict the shelf-life of hurood cheese rapidly and nondestructively, near infrared (NIR) spectral data of vacuum-packed hurood cheese at different storage times was obtained using a portable near-infrared spectrometer (from 950 to 1650 nm). The moisture, protein, coliform bacteria, mold and total bacteria count, were measured by traditional methods and the 12th day was assessed as the milestone for the shelf-life of hurood cheese. Spectral data was preprocessed separately by haar wavelet, db5 wavelet, sym5 wavelet, coif1 wavelet and bior1.1 wavelet soft threshold to denoise the spectra. 48 samples during shelf-life and 72 expired samples were divided into calibration set and validation set according to the ratio of 3:1 respectively. Then by using spectral data of calibration set and the corresponding class-labels (in shelf-life or outside shelf-life) the identification models of shelf-life period were established based on radial basis function neural networks (RBF-NN) while the validation set was for validating the models. Results showed that, RBF-NN with wavelet pretreatment was able to define the critical day during hurood cheese shelf-life. The portable NIR spectrometer excels in rapidly and nondestructively evaluating the quality and shelf-life of the vacuum-packed hurood cheese.


hurood cheese; NIR; shelf-life; RBF-NN; waveletText


Full Text:



  • There are currently no refbacks.