光致劣变大米的无损判别及不同遮挡方式有效性研究

目的 基于线性渐变分光便携式近红外光谱仪,建立光致劣变大米无损判别模型,并评估不同遮挡方式对大米光致劣变的预防效果。方法 以北方粳米、南方粳米、籼米、糯米、紫米、黑米为研究对象,分别各采用直接暴露于紫外光下、纸遮挡、白布遮挡、红布遮挡、玻璃遮挡、聚乙烯塑料遮挡方式制备实验组样品。采用线性渐变分光便携式近红外光谱仪采集大米样品近红外光谱。首先对6种大米分别建立单品种大米的光致劣变判别分析模型并优化数据预处理和光谱波段;然后将6种大米合并一起建立判别模型;最后采用优化模型预测各种遮挡的大米样品,以评估不同遮挡方式对大米光致劣变的预防效果。结果 基于全谱数据的标准正态变量结合去趋势校正预处理,所建判别模型的效果最佳, 6种大米判别模型的马修斯相关系数皆达到100.0%;而将6种大米一起建立判别模型,所建模型的预测性能有所下降。对每种大米采用各自优化模型对不同遮挡方式的实验组样品进行预测的结果表明,玻璃遮挡、聚乙烯塑料遮挡对大米光致劣变的预防有效率在97.0%以上,而纸遮挡、白布遮挡、红布遮挡在大米光致劣变的预防方面效果甚微。结论 基于近红外技术建立单品种大米光致劣变无损判别模型具有较高的预测...

Study on the non-destructive discriminant of photoinduced deterioration rice and the effectiveness of different shielding methods

Objective To develop the non-destructive discriminant models for photoinduced deterioration rice based on linear variable filter portable near-infrared spectrometer and assess the preventive effects of different shielding methods on photoinduced deterioration of rice. Methods In this study, the northern japonica rice, southern japonica rice, indica rice, glutinous rice, purple rice, and black rice were taken as the research objects, and the experimental group samples were prepared by direct exposure to ultraviolet light, paper shielding, white cloth shielding, red cloth shielding, glass shielding, and polyethylene plastic shielding, for each variety of rice respectively. The near-infrared spectra of the rice samples were collected by linear variable filter portable near-infrared spectrometer. First of all, the discriminant analysis models of photoinduced deterioration for each of the 6 varieties of rice were developed, and the data preprocessing and spectral bands were optimized. In addition, the discriminant analysis model for the 6 varieties of rice was developed together. Finally, the shielding rice samples were predicted to assess the preventive effect of different shielding methods on photoinduced deterioration of rice. Results It was demonstrated according to the research that the partial least square discriminant analysis models developed based on the data preprocessing of standard normal variate combined with detrend correction with full band spectral data are of the best performances for each of the 6 varieties of rice, with the Matthews correlation coefficients of 100.0% for the discriminant models of the 6 varieties of rice. However, the prediction accuracy of the partial least square discriminant analysis model developed by all the 6 varieties of rice has been decreased. It was indicated from the prediction results of experimental group samples with different shielding methods by the optimization models for the 6 varieties of rice respectively that the glass shielding and the polyethylene plastic shielding had an effective rate of over 97.0% in preventing photoinduced deterioration of rice, while paper shielding, white cloth shielding, and red cloth shielding had little effect in preventing photoinduced deterioration of rice. Conclusion The development of the non-destructive discriminant model for photoinduced deterioration of single variety of rice based on near-infrared spectroscopy is of the high prediction accuracy, moreover, glass and polyethylene plastic can prevent the occurrence of photoinduced deterioration of rice effectively.