近红外光谱模型转移新算法

发布时间：2018-03-22 03:08

本文选题：近红外光谱　切入点：化学计量学　出处：《化学进展》2017年08期 　论文类型：期刊论文

【摘要】：模型转移是解决分析仪器或分析方法通用性的关键技术。近红外光谱受测量仪器或测量条件的影响较大,模型转移对近红外光谱技术的实际应用尤为重要。本文综述了近年来近红外光谱分析中被广泛应用和新提出的模型转移算法,从计算原理角度梳理了有标样和无标样算法的联系和区别。有标样算法重点介绍了基于多元校正、因子分析、人工神经网络、多任务学习的模型转移方法,无标样算法重点介绍了基于光谱校正、模型参数校正和稳健建模的模型转移方法。从算法的角度分析了各种模型转移方法的特点和转移效果,并展望了模型转移算法的进一步发展。在综述的众多方法中分段直接标准化及其变体仍是模型转移的黄金标准,但是,基于因子分析的算法正变得受欢迎且基于神经网络和多任务学习的方法近年来也吸引了越来越多的注意。但是,在实际应用中,获得标准样品以在主机和子机上测得其光谱比较困难甚至是不可能的,无标样模型转移则更加实用。此外,随着仪器小型化、成像及超光谱成像的发展,模型转移在未来会变得愈加必不可少。
[Abstract]:Model transfer is the key technology to solve the universality of analytical instruments or analytical methods. Near-infrared spectroscopy is greatly affected by measuring instruments or measuring conditions. Model transfer is particularly important for the practical application of near infrared spectroscopy. In this paper, the widely used near infrared spectroscopy analysis and the new model transfer algorithm are reviewed. From the point of view of calculation principle, this paper combs the connection and difference between standard and non-standard algorithm. The algorithm of standard sample mainly introduces the model transfer method based on multivariate correction, factor analysis, artificial neural network and multitask learning, which is based on multivariate correction, factor analysis, artificial neural network and multi-task learning. The model transfer method based on spectral correction, model parameter correction and robust modeling is introduced. The characteristics and effects of various model transfer methods are analyzed from the point of view of the algorithm. The paper also looks forward to the further development of model transfer algorithms. Among the many methods reviewed, direct standardization and its variants are still the gold standard of model transfer. The algorithm based on factor analysis is becoming popular and the methods based on neural network and multitask learning have attracted more and more attention in recent years. It is difficult or even impossible to obtain standard samples to measure their spectra on mainframe and subcomputer, and it is more practical to transfer the model without standard. In addition, with the development of instrument miniaturization, imaging and hyperspectral imaging, Model shift will become increasingly necessary in the future.
【作者单位】：南开大学化学学院分析科学研究中心;天津市生物传感与分子识别重点实验室;药物化学生物学国家重点实验室;天津化学化工协同创新中心;喀什大学化学与环境科学学院;
【基金】：国家自然科学基金项目(No.21475068)资助~~
【分类号】：O657.33

【相似文献】