基于拉曼光谱高斯分峰拟合的硫化锑含量检测方法  

A Novel Method for Measuring Antimony Sulfide Content Based on Gaussian-Peak Fitting of Raman Spectroscopy

作　　者：李原鹰 徐德刚[1] 桂卫华[1] 阳春华[1] 蔡耀仪[1] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《光谱学与光谱分析》2017年第12期3743-3748,共6页Spectroscopy and Spectral Analysis

基　　金：国家自然科学基金项目(61473319);国家自然科学基金重点项目(61533021);国家自然科学基金创新研究群体项目(61621062);中南大学创新驱动计划资助

摘　　要：针对硫化锑含量采用化学分析方法检测存在操作复杂、检测时间长的问题,提出了一种基于拉曼光谱高斯分峰拟合的硫化锑含量快速检测方法。采用拉曼光谱系统测定锑矿样品的拉曼谱图,对原始拉曼谱图进行平滑去噪、背景扣除、谱段选择及归一化等预处理。基于高斯曲线的高斯峰、峰面积、半高宽和峰位置等信息,建立单个高斯峰的数学模型,提出高斯分峰拟合算法对光谱进行拟合,采用状态转移算法对高斯峰模型进行优化求解,得到表征光谱信息的特征参量,结合偏最小二乘回归方法,确定特征参量和硫化锑含量之间的关系,从而建立模型,实现对硫化锑含量的预测。实验中通过训练样本建立校正模型,对测试样本进行预测,同时从训练样本中随机挑选出检验样本,利用已建立的模型,对其硫化锑含量进行预测,以检验模型的正确性和外推性。实验结果表明:与预处理后全光谱建模相比,采用高斯分峰拟合后建立的预测模型的预测效果更好,证明了模型的正确性和良好外推性。因此,拉曼光谱结合高斯分峰拟合算法应用于锑矿中硫化锑含量的检测是可行的,且测量过程更简单,适用于矿物成分的快速分析。For measuring antimony sulfide by artificial chemical methods had many problems ,such as complex operation and long detection time ,a novel method for measuring antimony sulfide content rapidly based on Gaussian-peak fitting of Raman spectros-copy was proposed .The antimony samples were firstly characterized by Raman spectroscopy system to get their Raman spectros-copy ,and Raw Raman spectra were pretreated including smoothing ,background subtraction ,spectrum selection and normaliza-tion .The mathematical model of single Gauss peak was established based on Gaussian peaks ,peak area ,half high width and peak position of the gauss curve .Then the Gaussian-peak fitting algorithm was proposed to fit Pretreated Raman spectra .The state transition algorithm was employed to get the optimization solution .The characteristic parameters were obtained to describe the information of Pretreated Raman spectra .The relational model between the spectra and antimony sulfide content was estab-lished by the combination of the Gaussian characteristic parameters and partial least squares regression method ,so as to predict antimony sulfide content .It was important to verify this model .In the experiment ,the calibration model was established by the training samples and was used to predict the testing samples .In order to test the correctness and extrapolation of the model ,the proving samples were randomly selected from the training samples ,and then the established model was used to predict the anti-mony content of them .The experimental results showed that applying Gaussian-peak fitting of Raman spectroscopy to measure antimony sulfide content was feasible ,and the measuring process was more simple .It is suitable for rapid analysis of mineral composition .

关 键 词：拉曼光谱 高斯分峰拟合 硫化锑含量 优化求解 检测 

分 类 号：O657.3[理学—分析化学]