近红外分析中光谱波长选择方法进展与应用___

光谱波长选择方法在近红外光谱分析技术中相当重要。在复杂的光谱信息中剔除冗余信息、提取有用信息,可以提高光谱分析定量校正模型的预测精度和建模效率,得到预测能力强、稳健性好的近红外校正模型。本文综述了目前常用于偏最小二乘方法(PLS)建模的近红外波长选取方法及这一领域的最新进展,详细介绍遗传算法(GA)、间隔偏最小二乘方法(IPLS)等波长选取方法,并给出了这些方法的一些应用实例。

In the past decade,near infrared(NIR) spectral analysis technique has been quickly developed and widely applied in virtue of the development of chemometrics,in which spectral wavelength selection methods play an important role.Discarding irrelevant information and extracting essential information in complex spectral information can improve the spectral analysis of quantitative calibration model prediction precise and modeling efficiency,which is helpful to construct a good robustness NIR calibration model with strong forecasting capabilities.In the paper,the typical and commonly used wavelength selection methods are described.Some newly developed methods in this field such as genetic algorithm(GA) and interval partial least squares(IPLS) methods are introduced in detail.The algorithms and applications in NIR analysis of those methods are given and discussed.