随着经济水平的提升人们对大米品质要求越来越高, 由于不同大米品种之间价格差异也较大, 致使不少商贩以劣充优谋取利益, 有的掺和比例高达30%以上, 这种行为严重损害了消费者利益。 大米作为一种碳水化合物直接通过一维近红外光谱信息不易区分掺和米, 目前诸多研究集中在基于一维光谱的化学计量学判别模型建立。 二维相关光谱具有高分辨率、 解析峰的归属等优点, 可以挖掘出掺和米在一维光谱中隐藏的有效信息。 以五常大米作为研究对象, 选取难以用肉眼分辨的六种大米为掺入米, 分别制备5%~50%的不同掺和比例大米样品140个。 以五常大米近红外光谱的平均光谱作为参考谱, 掺和比例作为外部扰动因数, 将掺和米光谱和五常米光谱分别与参考谱进行二维相关运算, 通过解析不同掺和比例大米二维相关同步谱特性发现自相关谱1 420和1 920 nm两处自动峰值与同步谱(1 420, 1 920) nm和(1 920, 1 420) nm处交叉峰值强度均随掺和比例增加呈递增趋势, 其中1 920 nm自动峰值对掺和比例响应最显著。 通过对自相关谱1 420和1 920 nm两处自动峰产生机制的追溯并分析对应官能团归属, 发现大米中直链淀粉对掺和比例响应程度高于蛋白质及其他碳水化合物。 分别以五常大米同步谱中1 420和1 920 nm处自动峰值和(1 920, 1 420)nm处交叉峰值的最大值作为判别阈值, 对140个大米样品进行了判别试验。 结果显示, 基于1 920 nm自动峰值的判别效果最佳正确率达93.3%, 掺和比例20%及以上掺和米样品判别正确率为100%, 随着掺和比例降低判别正确率也逐渐下降, 掺和比例15%, 10%和5%样品判别正确率分别为91.7%, 66%和75%。 综上所述, 以掺和比例作为外部干扰因数解析不同掺和比例的大米二维同步谱特性, 通过特征峰值的差异可以简单有效区分掺和米, 与以往近红外判别模式相比不需要准备大量样品来训练模型, 为快速鉴别掺和大米提供一种新思路。

With the improvement of economic level, people have higher and higher requirements on the quality of rice. Due to the large price difference among different rice varieties, many traders seek profits by filling the inferior with the superior, sometimes the adulteration ratio is as high as 30%, which seriously damages the interests of consumers. Rice as a carbohydrate, which cannot be distinguished from adulterated rice by one-dimensional near-infrared spectroscopy. At present, many kinds researches focus on the establishment of stoichiometric discrimination model based on one-dimensional spectrum. Two-dimensional correlation spectra have the advantages of high resolution, and the attribution of analytical peaks and the effective information hidden in the one-dimensional spectrum of adulterated rice can be further explored. This paper takes Wuchang rice as the research object, selects six kinds of rice which are difficult to be distinguished by the naked eye as the adulterated rice, and prepares 140 rice samples with a different adulterated proportion of 5% to 50% respectively. The mean spectrum of Wuchang rice was taken as the reference spectrum and the mixing ratio as the external disturbance factor, the NIR spectra of adulterated and Wuchang rice were calculated with the reference spectrum. By analyzing the characteristics of Two-dimensional correlated synchronous spectra of rice with different mixing ratios, it was found that the cross peak intensity of auto-correlation spectra at 1 420 and 1 920 nm and synchronous spectra at (1 420,1 920) and (1 920, 1 420) nm increased with the increase of adulteration ratio, and the 1 920 nm automatic peak has the most significant response to the adulteration ratio. By tracing the generation mechanism of two automatic peaks at 1 420 and 1 920 nm of the autocorrelation spectrum and analyzing the attribution of corresponding functional groups, it was found that the response degree of amylose in rice to the adulteration ratio was higher than that of protein and other carbohydrates. The maximum value of the automatic peak at 1 420 and 1 920 nm and the maximum value of the cross peak at (1 920,1 420) nm in the Wuchang rice synchronous spectrum were used as the discriminant threshold to discriminate 140 rice samples. The results showed that the best discriminant accuracy was 93.3% based on the 1920nm automatic peak value, and the discriminant accuracy was 100% for the adulterated rice samples with the adulterated ratio of 20% or more, As the adulterated ratio to reduce the discriminant accuracy also gradually decline, with 15%, 10%, 5% sample discriminant accuracy were 91.7%, 66% and 75% respectively. To sum up, the blending ratio as the external interference factor and the characteristics of the two-dimensional synchronous spectrum of rice with different adulteration ratios were analyzed, the adulterated rice can be distinguished simply and effectively by the difference of characteristic peak value, compared with previous NIR discriminant models, it is not necessary to prepare a large number of samples to train the model, which provides a new idea for the rapid identification of adulterated rice.