论文部分内容阅读
为对劣变花生进行快速识别,本文采用太赫兹波谱仪结合衰减全反射附件(ATR)采集了劣变花生、非劣变花生的太赫兹波谱数据。在比较太赫兹谱形的基础上,对所采集的太赫兹吸收系数谱数据两两之间计算相关系数,研究相关系数的分布特征,并进一步建立劣变花生太赫兹吸收系数谱的偏最小二乘-判别分析(PLS-DA)模型。分析结果表明,经过一阶导数处理,劣变花生、非劣变花生分别的相关系数的样本标准差更大,说明一阶导数可较明显地体现花生太赫兹波谱数据的差异;小样本容量的PLS-DA模型正确率可达到100%,劣变花生、非劣变花生的交互验证正确率分别为80%、90%。研究表明,太赫兹波谱在劣变花生快速鉴别方面具有一定的应用潜力,亦可为其他劣变农产品的快速识别提供一定的参考。
In order to identify the bad peanut rapidly, the terahertz spectroscopy data of the bad peanut and non-inferior peanut were collected by using the THz spectrometer and attenuated total reflection attachment (ATR). Based on the comparison of terahertz spectra, the correlation coefficients were calculated between the two pairs of terahertz absorption spectrum data, the distribution characteristics of the correlation coefficients were studied, and the partial minimum of the terahertz absorption coefficient spectrum Multiply-Discriminant Analysis (PLS-DA) model. The results show that after the first derivative treatment, the standard deviation of the correlation coefficient of the bad peanut and the non-bad peanut is larger, which shows that the first derivative can obviously reflect the difference of the peanut terahertz spectrum data; the small sample capacity The correct rate of PLS-DA model can reach 100%, and the correct rate of mutual validation of bad peanut and non-bad peanut is 80% and 90% respectively. Studies have shown that terahertz spectroscopy has potential applications in the rapid identification of bad peanuts and may provide some references for the rapid identification of other degraded agricultural products.