组合生成算法与多元线性回归相结合用于近红外光谱波长的优选

分立波长型近红外光谱分析仪是光谱分析仪器中较为普及的一种快速成份定量分析仪。如滤光片型、发光二极管型等。该类分析仪器研发的一个主要问题是如何针对于待测物质主要成份进行近红外光谱解析。找到最优定标波长组合用于建立稳健的定标模型。常用的波长选择方法为相关光谱结合逐步多元线性回归方法，该方法依据各参与定标波长所对应的t检验值进行最优定标波长的判别，但在实际应用中定标模型的定标精度和预测精度相差较大，具有很大的不准确性。为了实现定标波长的优选引入了组合数学中的组合生成算法。可以在较短的时间内完成最优波长组合的选取，结果是令人满意的。

Choose Optimal Wavelengths for Calibration by Combining the Combination-making Algorithm with Multivariate Linear Regression

Fixed filter spectrometer, diode arrays spectrometer and so on are very common spectrometers which don't operate in the consistently scan mode. The key to this kind of spectrometer is how to find a optimal wavelength combination rapidly and accurately for getting a fine calibration. The conventional methods to choose the optimal wavelengths for calibration are forward stepwise multiple linear regression in combination with correlation chart, which is based on the t test result of all the calibration wavelengths. However, the wavelengths chosen by conventional method are sometimes not so powerful, as a result it contains uncertainty and is undependable. In this paper the combination making algorithm in combinatorics is introduced in near infrared spectroscopy, the work of finding out the optimum wavelength combination is realized efficiently and successfully, the final result of prediction of the optimum calibration model is satisfied.