为了预测鲜枣常温贮藏的保鲜期,确保鲜枣的品质要求及食用安全,应用近红外光谱建立了室温贮藏下鲜枣内部霉菌菌落总数变化的动力学模型.通过对几种数据预处理方法的比较及特征波数的选择,实现了鲜枣霉菌菌落总数变化的近红外模型的优选.结果表明：经过多元散射校正处理的鲜枣近红外光谱,应用多元线性回归方法建立的霉菌菌落总数模型预测能力较好,校正集相关系数为0.920,均方根误差为1.503,预测集相关系数为0.889,均方根误差为1.514.同时,将近红外光谱模型应用于霉菌菌落总数随贮藏时间变化的零级反应动力学模型中,得到模型的相关系数为0.981.根据近红外光谱吸光度值与贮藏时间的线性关系,当霉菌菌落总数初始值小于等于10 cfu/g 时,预测出鲜枣在室温下的保鲜期一般为8 d.研究表明,结合动力学模型的近红外光谱技术可以作为一种无损、快速检测方法来检测鲜枣霉菌菌落总数变化.%The objectives of this study were: 1) to optimize a near-infrared (NIR) spectroscopy model for fresh jujube to predict the quality deterioration (mold growth) during storage at room temperature, 2) to establish a kinetic model of mold growth according to NIR spectroscopy and storage time at room temperature, and 3) to predict the shelf life of fresh jujube at room temperature. The 72 Lizao samples were picked and divided into 18 sets. The mold infection level was measured, and the NIR spectroscopy was obtained from 4 samples which were tested every day. The spectral data of each jujube measured at three locations were first averaged. The samples were then divided into calibration sets and prediction sets for spectral data analysis. In order to optimize the NIR model, the pretreatment techniques such as Savitzky-Golay smoothing (S-G smoothing), multiplicative scatter correction (MSC), first derivative (1-Der) and second derivative (2-Der) were compared with the raw spectra by using a statistical software package (Unscrambler 9.8), and the regression coefficient (RC) method was used to determine the characteristic wave number. Multiple linear regression (MLR) was applied as NIR modeling method. According to the predicted mold infection level using NIR model, the chemical kinetic models of spectral data and storage time at room temperature with zero-order and first-order reaction were established by using a statistical software package (SPSS 18). The shelf life could be predicted based on these chemical kinetic models. The results showed that the characteristic wavenumber of 10 300, 8 330, 6 900, 5 666, 5 150 and 4 060 cm-1 in the near-infrared range with MSC technique could be chosen to model the quality deterioration of fresh jujube at room temperature. The NIR model that produced the best prediction had the form: A=128.812−115.680X1−150.692X2−19.312X3+15.821X4+4.345X5−0.320X6, where A is mold value (lg/cfu·g-1), X1–X6 are absorbance value of characteristic wavenumber. The correlation coefficient of calibration (Rc) was 0.920, the root mean square error of calibration (RMSEC) was 1.503, the correlation coefficient of prediction (Rp) was 0.889, the root mean square error of prediction (RMSEP) was 1.514. The zero-order reaction kinetic model performed better than the first-order model. The zero-order reaction kinetic model of mold growth with storage time was predicted by At=0.690−0.020 (128.812−115.680X1−150.692X2−19.312X3+15.821X4+4.345X5−0.320X6) +0.301t, with a correlation coefficient of 0.981. Based on the linear correlation between the NIR measurement and storage time, the shelf life of fresh jujube at room temperature was predicted to be 8 days with the mold infection level less than 10 cfu/g. The calculation of MLR method is easier than the partial least squares (PLS) method and principal component regression (PCR) method. The MLR method was more suitable for developing the low cost portable spectrum instrument. The variety of fresh samples used in this test was Lizao at room temperature, and further tests should be conducted to verify the model for its applicability to other varieties. The study showed that the NIR combed with kinetic models could be used to predict the shelf life and ensure the quality and safety of fresh jujube.
展开▼
