LIBS光谱数据分类算法及应用研究
发布时间：2018-08-05 19:24
【摘要】：激光诱导击穿光谱技术(Laser-induced breakdown spectroscopy, LIBS)是一种新型的元素分析技术,具有实时在线、非接触、多种元素同时探测等优点,是光谱分析领域的一种前沿性分析手段,在石油、冶金、地质、环保、军事、航天等领域都有广泛的应用。但由于其昂贵的成本,核心技术的垄断以及可能涉及到的重要战略作用,使其引进受限。因此研制开发功能更加完善,具备自主知识产权,并且成本较为低廉的创新型激光光谱仪器具有重要的理论意义和现实意义。 论文在综述LIBS光谱数据分类算法的基础上,对光谱预处理、光谱定性和定量分析以及光谱识别与解析等方面进行优化设计、算法实现及软件集成。主要研究工作包括：(1)针对LIBS光谱数据特点,引入鲁棒的主成份分析算法(Robust Principal Component Analysis, RPCA)对光谱数据滤噪和降维,并利用支持向量机(Support Vector Machine,SVM)分类模型进行分类,提高了预测准确率,通过对比实验验证了方法的有效性和可行性；(2)针对单一分类器分类效果不稳定的缺陷,通过AdaBoost组合分类模型将偏最小二乘法和支持向量机混合构建分类模型,实现了9种牌号圆钢的有效分类,实验验证了文中算法可大大提高模型分类正确率和分类模型的稳定性；(3)最终设计并开发了集谱图解析、定性与定量分析、识别与感知功能于一体的“LIBS光谱预处理与分析系统”。 研究结果可应用于光谱数据分类的研究,并为创新型激光光谱仪器的数据分析与处理提供所需的支持。
[Abstract]:Laser induced breakdown spectroscopy (Laser-induced breakdown spectroscopy, LIBS) is a new element analysis technology, which has the advantages of real-time on-line, non-contact, simultaneous detection of multiple elements and so on. It is a leading analytical method in the field of spectral analysis, which is used in petroleum, metallurgy, and so on. Geology, environmental protection, military, aerospace and other fields have been widely used. However, due to its high cost, monopoly of core technology and the important strategic role that may be involved, its import is limited. Therefore, it is of great theoretical and practical significance to develop innovative laser spectrometer with more perfect functions, independent intellectual property rights and low cost. On the basis of summarizing the classification algorithms of LIBS spectral data, this paper optimizes the design, algorithm realization and software integration of spectral preprocessing, spectral qualitative and quantitative analysis, spectral identification and analysis. The main research work includes: (1) according to the characteristics of LIBS spectral data, a robust principal component analysis (Robust Principal Component Analysis, RPCA) algorithm is introduced to filter noise and reduce the dimension of spectral data, and the support vector machine (SVM) classification model is used to classify the spectral data, which improves the prediction accuracy. The effectiveness and feasibility of the method are verified by comparative experiments. (2) aiming at the unstable effect of single classifier, the partial least square method and support vector machine are combined to construct the classification model by AdaBoost combined classification model. The experimental results show that the algorithm can greatly improve the classification accuracy and stability of the model. (3) finally, we design and develop the set spectrum analysis, qualitative and quantitative analysis. LIBS spectral preprocessing and analysis system with recognition and sensing functions. The results can be applied to the classification of spectral data and provide the necessary support for the data analysis and processing of the innovative laser spectrometer.
【学位授予单位】：西北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.4;TP301.6
本文编号：2166760
[Abstract]:Laser induced breakdown spectroscopy (Laser-induced breakdown spectroscopy, LIBS) is a new element analysis technology, which has the advantages of real-time on-line, non-contact, simultaneous detection of multiple elements and so on. It is a leading analytical method in the field of spectral analysis, which is used in petroleum, metallurgy, and so on. Geology, environmental protection, military, aerospace and other fields have been widely used. However, due to its high cost, monopoly of core technology and the important strategic role that may be involved, its import is limited. Therefore, it is of great theoretical and practical significance to develop innovative laser spectrometer with more perfect functions, independent intellectual property rights and low cost. On the basis of summarizing the classification algorithms of LIBS spectral data, this paper optimizes the design, algorithm realization and software integration of spectral preprocessing, spectral qualitative and quantitative analysis, spectral identification and analysis. The main research work includes: (1) according to the characteristics of LIBS spectral data, a robust principal component analysis (Robust Principal Component Analysis, RPCA) algorithm is introduced to filter noise and reduce the dimension of spectral data, and the support vector machine (SVM) classification model is used to classify the spectral data, which improves the prediction accuracy. The effectiveness and feasibility of the method are verified by comparative experiments. (2) aiming at the unstable effect of single classifier, the partial least square method and support vector machine are combined to construct the classification model by AdaBoost combined classification model. The experimental results show that the algorithm can greatly improve the classification accuracy and stability of the model. (3) finally, we design and develop the set spectrum analysis, qualitative and quantitative analysis. LIBS spectral preprocessing and analysis system with recognition and sensing functions. The results can be applied to the classification of spectral data and provide the necessary support for the data analysis and processing of the innovative laser spectrometer.
【学位授予单位】：西北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.4;TP301.6
【参考文献】
相关期刊论文 前3条
1 刘凯;王茜劏;赵华;肖银龙;;激光诱导击穿光谱在塑料分类中的应用[J];光谱学与光谱分析;2011年05期
2 赵华;王茜劏;刘凯;葛聪慧;;无机爆炸物及其主要成分的激光诱导击穿光谱实验研究[J];光谱学与光谱分析;2012年03期
3 王茜劏;黄志文;刘凯;李文江;阎吉祥;;基于主成分分析和人工神经网络的激光诱导击穿光谱塑料分类识别方法研究[J];光谱学与光谱分析;2012年12期
相关博士学位论文 前1条
1 姚顺春;激光诱导击穿光谱技术在电站运行诊断中的应用研究[D];华南理工大学;2011年
，本文编号：2166760
本文链接：https://www.wllwen.com/falvlunwen/zhishichanquanfa/2166760.html
