CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleResearch on the Temperature-Induced Spectral Variation and Correction of the Temperature Effect
Thesis Advisor彭思龙
Degree Grantor中国科学院大学
Place of Conferral中国科学院自动化研究所
Degree Discipline模式识别与智能系统
Keyword红外光谱 温度预测 温度校正 多元校正模型 Infrared Spectroscopy Temperature Prediction Temperature Correction Multivariate Calibration
Abstract温度校正一直是光谱定量分析中的一个基础问题。红外光谱的温度校正预处理,目的是去除光谱数据中由于温度波动所带来的非线性效应,从而提高光谱数据对于浓度的线性度。光谱定量模型的线性度得到改善后,模型的预测精度也会得到相应的提升。通常的温度校正方法不仅需要记录训练集光谱的采集温度,而且需要记录测试集光谱的采集温度,这对很多实际应用中的光谱温度校正造成了困难。本文提出了一种基于模型的光谱温度预测及校正方法,通过训练集数据对光谱中的温度信息进行建模,利用模型的信息,从而能从测试集光谱数据中估计出采集温度,并进行光谱数据的温度校正,降低了温度校正方法对测试集光谱数据采集温度的依赖性,从而大大提升了温度校正在红外光谱定量分析中的适用性。 作为方法的验证,本文进行了三部分的实验:在第一部分的实验中,我们通过对十个浓度的水-乙醇二元混合物光谱数据的温度预测以及温度校正的实验,证明了本文方法的有效性;在第二部分的实验中,我们采用了Wulfert的经典温度校正方法CPDS的实验数据和实验方案,对三元混合物的光谱数据进行温度预测以及温度校正,得到了不亚于CPDS方法的温度校正效果,同时也证明了本文方法对三元混合物光谱数据的适用性;在第三部分的实验中,我们结合本文模型和经典的PLS模型的浓度预测结果,提出了一系列迭代温度校正算法,说明了本文提出的温度模型和温度校正方法具有很强的可扩展性。三部分的实验结果表明,在缺少测试集测量温度的情况下,本文提出的温度校正方法仍可对光谱数据进行有效的预测和校正,降低了温度校正方法对测试集数据的依赖性,从而提高了温度校正方法的适用性。
Other AbstractTemperature correction of infrared spectroscopy, aiming at removing nonlinear temperature effects on the spectra, has been a basic problem in Infrared Spectroscopy Quantitative Analysis for a long time. By improving the linearity of spectrum, we can get a better prediction result from multivariate calibration models such as PLS. Previous temperature correction methods need to record all the collecting temperatures of the spectra, not only for those in the training set, but also for those in the test set, which brings a lot of difficulties to real application. In this paper, we propose a new model-based method for temperature prediction and temperature correction of infrared spectral data. It works by modeling the temperature effect in the spectrum, so that we can get a predicted collecting temperature of the spectrum in the test set by its variations. The predicted collecting temperature is then regarded as the real collecting temperature. In this paper, we carry out three experiments to support our theory. In the first part, we test the temperature prediction results and temperature correction results of some spectra, which are collected from the water-ethanol binary mixtures in the test set. The results show that our method is effective and efficient. In the second part, we compare our method with CPDS, a classic temperature correction method proposed by Wulfert. We repeat the experiment design in his experiment section, using our method and his data set. We carry out the temperature correction experiment in this paper and then compare the temperature correction results of our method with his. Results show that, despite the lack of collecting temperatures of the test set spectra, our method can also offer a satisfactory result of temperature correction as the classic CPDS method. In the last part, we combine the predicted concentrations of our model and those of PLS model, and then propose a series of concentration prediction methods based on iterative algorithm. The three experiments show that, despite the absence of measured collecting temperatures of the spectra in the test set, our method proposed in this paper is still effective and efficient to remove the temperature effect in the spectra, thereby making temperature correction easier to use in real application.
Other Identifier201228014628058
Document Type学位论文
Recommended Citation
GB/T 7714
杨驰晓. 光谱信号温度效应的研究与去除[D]. 中国科学院自动化研究所. 中国科学院大学,2015.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20122801462805(1453KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[杨驰晓]'s Articles
Baidu academic
Similar articles in Baidu academic
[杨驰晓]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[杨驰晓]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.