论文部分内容阅读
面对近红外应用模型共享较难问题,研究采用shenk’s算法解决烟叶化学成分近红外光谱分析模型应用中的模型转移。为实现烟叶近红外光谱模型转移,以山东、云南、四川、广西四个产区的烟叶作为分析对象,采用Thermofisher Nicolet AntrisII相同型号的两台仪器采集近红外光谱分别作为主机和子机,在主机上建模,用shenk’s和斜率/截距两种转移算法对子机进行模型转移对比分析。同时,使用ICCA国际标准和t检验两种方法评价模型的传递性。研究结果表明,利用shenk’s算法能够从根本上对模型进行转移,满足模型传递误差要求,且预测结果与主机不存在显著性差异,斜率/截距法没达到稳定的效果,所以shenk’s算法可有效的实现烟叶模型传递问题。
In the face of the difficulty of near-infrared application model sharing, shenk’s algorithm is used to solve the model transfer in the application of near-infrared spectral analysis of tobacco chemical composition. In order to achieve the transfer of near-infrared spectral model of tobacco leaves, the tobacco leaves from four producing areas of Shandong, Yunnan, Sichuan and Guangxi were used as the analysis objects. Two instruments of the same type of Thermofisher Nicolet AntrisII were used to collect near-infrared spectroscopy as host and sub-machine respectively. Modeling, using shenk’s and slope / intercept two transfer algorithm for the transfer of sub-machine model comparative analysis. At the same time, the ICCA international standard and t test were used to evaluate the transmissibility of the model. The results show that the shenk’s algorithm can fundamentally transfer the model to meet the model transfer error requirements, and the prediction results do not have significant differences with the host, the slope / intercept method does not achieve a stable effect, so shenk’s algorithm can be effective Achieve tobacco model transfer problem.