尚光网 本站首页 本所首页 联系我们
chi20181129上海光机所瞿荣辉557455Raman spectroscopy is a kind of spectral analysis technology based on Raman spectroscopy for the identification and content measurement of material components. It has the advantages of non-destructive, informative and sample preparation, and has become a hot research topic in recent years. The miniature Raman spectrometer has the characteristics of small size, simple operation and excellent performance, which makes the Raman spectroscopy technology gradually out of the laboratory and has been widely used in many fields such as food safety, drug testing and jewelry identification. In order to obtain quasi-deterministic analysis results of substances by Raman spectroscopy under non-manual operation, automatic and rapid realization of Raman spectroscopy data processing is the top priority of Raman spectroscopy qualitative analysis technology. Most of the current Raman spectroscopy studies focus on the analysis and matching of pure Raman spectroscopy, and lack of research on the mixture Raman spectroscopy. The most important feature of the mixture Raman spectroscopy is that there may be a large number of overlapping peaks in the Raman spectrum due to the similarity of the molecular structure between the components of the analyte, resulting in the pure Raman spectral recognition technique not applicable. The identification of the components of the mixture is carried out, especially for Raman spectrometers with low spectral resolution, the presence of overlapping peaks makes the identification of the components of the mixture more difficult. At present, spectral matching algorithms based on library search are more suitable for pure Raman spectroscopy. Therefore, how to achieve effective resolution of overlapping peaks and feature extraction of spectral peaks becomes a major problem in mixture Raman spectroscopy. At the same time, it is necessary to study a new algorithm for spectral matching of mixtures to achieve automatic and accurate identification of the components of the mixture. In this paper, the problems faced by Raman spectroscopy in the automatic identification of mixture components are studied theoretically and experimentally in Raman spectral peak identification, overlapping peak resolution and feature extraction, and spectral library-based mixture spectral matching algorithm. The research contents are summarized as follows: 1) Based on the analysis of Raman spectroscopy and its mathematical model, an overlapping peak resolution method combining spectral deconvolution algorithm and Lorentz4 wavelet transform algorithm is proposed. Compared with the traditional Mexican hat wavelet and the Gaus4 wavelet, the method has better overlapping peak resolution and improves the peak resolution to 0.4. Through the spectral peak-finding experiment of amino acid mixture, it is verified that the method has smaller false positive rate and higher peak recognition rate, so it has more stable peak identification performance. 2) A matching algorithm for the integrated spectral features of the mixture is proposed. The method constructs the integrated spectral feature matching coefficient by logistic regression mathematical model, fusion peak matching coefficient, non-negative least squares matching coefficient and common cosine matching coefficient in pure object matching. The matching algorithm not only takes into account the characteristic peak information of the mixture, but also takes into account the information contained in the full spectrum. More comprehensive information is referenced in the matching process compared to existing methods. Through the spectral matching experiment of amino acid mixture, it is verified that the method has better matching effect than the other three methods, and the component identification of amino acid mixture can be realized without misjudgment. This method provides new ideas and solutions for mixture matching, which is beneficial to improve the recognition ability of the components of the mixture. 3) A new Raman peak identification algorithm based on Voigt function fitting is proposed. According to the physical nature of Raman spectroscopy, Voigt function is used as a fitting function to fit the peaks obtained after peak finding of wavelet transform ridges. Processing, the peak similarity before and after the fitting is used as the basis for discriminating the peak, and the small amplitude Raman signal similar to the noise amplitude can be effectively identified. By comparing with the existing two spectral peak discrimination methods, the recognition rate of the algorithm is improved by 60% in the case of false positive rate of 5%; the false positive rate of the algorithm is obtained when the recognition rate is higher than 90%. Reduced by 10%. A Raman spectral background subtraction algorithm with unique parameters to be adjusted is improved. Compared with other multi-parameter adjustment methods, this method only needs to set the distance threshold parameter for controlling the background baseline smoothness, which is convenient for the parameter adjustment by the personnel of the spectrometer instrument, which is conducive to the development and promotion of portable Raman spectroscopy. 4) A multi-spectral diamond automatic identification instrument GEM CHKR based on Raman spectroscopy, photoluminescence spectroscopy and phosphorescence detection technology was developed. Raman spectroscopy is used to identify whether the object is a diamond, and the type of diamond is automatically identified by photoluminescence spectroscopy and phosphorescence detection. Compared to the existing diamond identification instruments on the market, GEM CHKR combines three spectral techniques to automatically identify true and pseudo diamonds and classify diamonds without the need for professional intervention, making the discriminating operation simpler and more standardized. By identifying Indian diamonds, the accuracy of natural diamonds and HPHT synthetic diamonds can reach about 90%. The instrument has important practical value for the rapid, accurate and batch identification of diamonds.2019atalunwen219121223597Raman spectroscopy, mixture, overlapping peak resolution, spectral library search, fluorescence background subtraction, diamond identificationResearch on Recognition Technology of Mixture Components Based on Raman Spectroscopy基于拉曼光谱的混合物组分识别技术研究拉曼光谱分析技术是一种基于拉曼光谱的对物质组分进行识别以及含量进行测量的光谱分析技术,其具有无损、信息丰富以及无需样品制备等优点,成为了近年来研究的热点。 微型拉曼光谱仪具有体积小巧、操作简单、性能优良等特点,使得拉曼光谱分析技术逐渐走出实验室,在食品安全、药物检测、珠宝鉴定等诸多领域得到了越来越广泛的应用。为了在非人工操作情况下,通过拉曼光谱获得对物质的准确定性分析结果,自动快速实现拉曼光谱数据处理是拉曼光谱定性分析技术的重中之重。 现阶段的拉曼光谱研究大多着眼于纯净物拉曼光谱的分析与匹配,缺乏对混合物拉曼光谱进行研究。混合物拉曼光谱的最大特点在于,由于被测物组分之间的分子结构可能存在相似之处,其拉曼光谱中可能存在大量的重叠谱峰,导致纯净物拉曼光谱识别技术不适用于进行混合物组分识别,尤其对于光谱分辨率低的拉曼光谱仪,重叠峰的存在使对于混合物组分的识别愈加困难。目前基于谱库检索的光谱匹配算法更多的是适用于纯净物拉曼光谱。因此,如何实现重叠峰的有效分辨和光谱谱峰的特征提取,成为混合物拉曼光谱分析的一大难题。同时需要研究针对混合物光谱匹配的新算法,用以实现对混合物的组分进行自动准确地识别。 本文针对拉曼光谱分析技术在混合物组分自动识别上面临的问题,在拉曼光谱谱峰识别、重叠峰的分辨和特征提取,以及基于谱库的混合物光谱匹配算法等方面进行理论和实验研究,现将研究内容总结如下: 1)通过对拉曼光谱产生机理以及其数学模型的分析和研究,提出了一种结合光谱解卷积算法与Lorentz4小波变换算法的重叠峰分辨方法。该方法相较于使用传统的Mexican hat小波以及Gaus4小波具有更佳的重叠峰分辨能力,将谱峰分离度提升至0.4。通过氨基酸混合物光谱寻峰实验,验证了该方法的误判率更小和谱峰识别率更高,因此具有更稳定的谱峰判别性能。? 2)提出了针对混合物的集成光谱特征的匹配算法。该方法通过逻辑回归数学模型,融合谱峰匹配系数、非负最小二乘匹配系数和纯净物匹配中常用的夹角余弦匹配系数,构建了集成光谱特征匹配系数。该匹配算法不仅考虑到混合物的特征峰信息,还考虑到了全谱所蕴含的信息。相较于现有的方法,在匹配过程中参考了更加全面的信息。通过氨基酸混合物光谱匹配实验,验证该方法相较于其它三种方法具有更好的匹配效果,能够实现氨基酸混合物的组分识别,没有出现误判的情况。该方法为混合物匹配提供了新的思路和解决方案,有利于提升对混合物组分的识别能力。 3)提出了一种新的基于Voigt函数拟合的拉曼谱峰判别算法,根据拉曼光谱的物理本质,采用Voigt函数作为拟合函数,对小波变换脊线寻峰后得到的谱峰进行拟合处理,将拟合前后的谱峰相似度作为判别谱峰的依据,可有效识别与噪声幅值相近的小幅值拉曼信号。通过与已有的两种谱峰判别方法进行对比,在误判率为5%的情况下,该算法识别率提高了60%;在识别率高于90%的情况下,该算法误判率降低了10%。改进了一种具有唯一需调参数的拉曼光谱背景扣除算法。该方法相较于其它多参数调节方法,只有一个用于控制拟合背景基线平滑度的距离阈值参数需要设置,方便使用光谱仪器的人员进行参数调节,有利于便携式拉曼光谱技术的发展和推广使用。 4)研制开发了基于拉曼光谱技术、光致发光光谱技术以及磷光检测技术的多光谱钻石自动识别仪GEM CHKR。通过拉曼光谱技术对被测物是否是钻石进行识别,通过光致发光光谱技术与磷光检测技术对钻石的种类进行自动识别。相比于市面上现有的钻石识别仪器,GEM CHKR融合了三种光谱技术,可以自动识别真、伪钻石和对钻石进行分类,无需专业人员干预判定,使得判别操作更加地简单和规范。通过对印度产钻石进行识别,对天然钻石和HPHT人工合成钻石的判别准确率可以达到90%左右。该仪器对于钻石的快速、准确、批量识别具有重要的实用价值。拉曼光谱分析技术,混合物,重叠峰分辨,光谱库检索,荧光背景扣除,钻石识别中国科学院上海光学精密机械研究所刘铭晖光学工程博士
中文题目: 基于拉曼光谱的混合物组分识别技术研究
外文题目: Research on Recognition Technology of Mixture Components Based on Raman Spectroscopy
作者: 刘铭晖
导师姓名: 瞿荣辉
学位授予机构: 中国科学院上海光学精密机械研究所
答辩时间: 20181129
中文关键词:
拉曼光谱分析技术,混合物,重叠峰分辨,光谱库检索,荧光背景扣除,钻石识别
英文关键词:
Raman spectroscopy, mixture, overlapping peak resolution, spectral library search, fluorescence background subtraction, diamond identification
中文摘要:
英文摘要:
文献类型:学位论文
学位级别: 博士
正文语种: chi
全文传递服务
clickdetails
页面点击量: 42
文章下载量: 1
visitlog
友情链接:
  上海大恒公司
  南京先进激光技术院
  光学产品库
版权所有 © 2009 中国科学院上海光学精密机械研究所 沪ICP备05015387号
主办:中国科学院上海光学精密机械研究所 上海市嘉定区清河路390号(201800)