LIPS技术及在农产品成分分析与种类识别中的应用初探

发布时间：2018-01-21 06:07

本文关键词： 激光诱导等离子体形成机制发射光谱激光诱导击穿光谱农产品　出处：《重庆大学》2016年博士论文　论文类型：学位论文

【摘要】：高功率脉冲激光聚焦到靶材目标上,不仅可以蒸发材料分子,同时也能产生明亮的发光等离子体。通过探测和分析这些等离子体的光谱信号,可以实现待测样品的定性与定量检测。因此,近年来,国内外科研人员提出了一种发展迅速的新型光谱分析技术——激光诱导等离子体光谱技术(Laser-Induced Plasma Spectroscopy,LIPS)。它因具有快速原位、实时在线、多元素同时检测等优点,在环境污染物监测、矿产勘测、爆炸物检测及深空探测等诸多领域的物质成分分析方面,具有广阔的应用前景。LIPS技术正逐渐成为基础研究和应用研究领域广泛关注的重要课题之一。目前,LIPS技术在金属样品的检测中已获得了成功应用。但对于包含多种金属、非金属元素成分的农产品,因其成分体系复杂、标样制备困难、成分分布不均,采用LIPS进行定量检测,尚存在着一些困难和挑战。激光在农产品表面的烧蚀机理比金属样品更复杂、所产生的双原子分子自由基CN和C原子辐射特性尚不明确;而且检测过程存在严重的基体效应,采用传统的标准曲线定标方法所获得的测量精度和线性范围较差。为探索LIPS技术在农产品检测领域的应用,本文首先对石墨的激光诱导等离子光谱特性和时间分辨特性进行了系统的研究,获得了LIPS检测单组份碳材料的基本规律;然后,结合LIPS与二元线性回归、BP神经网络两种定量方法,实现了对中药材独活根样品的定量分析;并利用LIPS技术结合支持向量机算法,实现了对6种茶叶样品的可靠分类和识别。本文的主要工作和结论如下:(1)开展了激光诱导石墨等离子体中CN(B2∑+?X2∑+)和C原子辐射光谱的时间演化特性研究。结果显示,CN(B2∑+?X2∑+)的发射持续时间是C原子的3倍,强度达到最大值的时间在0.2?s到0.8?s间。随着延迟时间的变化,电子温度从11807 K降低到8755 K,振动温度从8973 K降低到6472 K,转动温度从7288 K降低到4491 K。(2)研究了激光能量对激光诱导石墨等离子体中CN和C原子辐射的影响。结果显示,CN分子和C原子的激发阈值和光谱特性呈现不同特征:在低激光能量下,CN发射的电子激发温度、电子密度、振动温度和转动温度增加很快;在高激光能量下,CN发射的电子激发温度和电子密度增加较慢,由于等离子体屏蔽效应和CN分子的解离作用,振动温度和转动温度趋于饱和;而在同一时刻时的三种温度的大小关系为:TelecTvibTrot。实验测量了激光诱导石墨等离子体的电子密度,发现其电子密度在1017?1018 cm-3量级。(3)开展了中药材样品的LIPS检测技术研究。选取独活根样品中Al元素的308.22nm和309.27nm的两条特征光谱线作为分析线,建立二元线性回归方程,回归预测结果与ICP-AES得到的检测值结果拟合效果很好。这说明基于二元线性回归的定量方法可获得较好的测量精度,适用于LIPS技术检测农产品中的微量元素。(4)在独活根样品的LIPS光谱图中分别选取Ca、Mg和K元素的特征谱线Ca II 317.99 nm、Mg I 517.27 nm和K I 769.90 nm的光谱强度作为分析指标,绘制了Ca、Mg和K元素的定量分析的BP神经网络法的定标曲线,建立了中药材独活根样品中钙、镁和钾含量的光谱分析新方法,其中Ca、Mg和K的线性拟合系数R2分别达到98.96%、99.45%和99.11%。结果表明,利用BP神经网络法定量分析Ca、Mg和K具有更高的准确性和稳定性、更小的相对误差和相对标准偏差,验证了应用BP神经网络法对LIPS定量分析的可行性及优越性。(5)开展了LIPS技术对茶叶样品的检测研究,结合支持向量机算法成功实现了茶叶样品的分类判别。选取8条光谱线的相对强度作为分析指标建立了支持向量机判别模型。该模型对训练集的300个样品进行回判,对297个样本进行了正确的种类识别,识别正确率达99%。此外,对测试集中的300个样本进行了判别,对其中的291个样本进行了正确的种类识别,识别正确率达到97%。说明结合支持向量机算法的LIPS技术在农产品种类识别方面具有良好的应用前景。
[Abstract]:High power pulse laser to target the target, not only can the evaporation material molecules, also can produce plasma illuminating. Through the detection and analysis of spectral signals of these plasma, can achieve qualitative and quantitative detection of samples to be measured. Therefore, in recent years, researchers at home and abroad the research proposes a new spectrum of rapid development analysis of technology of laser-induced plasma spectroscopy (Laser-Induced Plasma Spectroscopy, LIPS). It is quick in situ, real-time online, multi-element detection etc. in the monitoring of pollutants, environmental mineral survey, material composition and explosives detection for deep space exploration and many other fields of analysis, with the application foreground of.LIPS it has gradually become one of the very important topic in the field of basic and applied research. At present, the LIPS technology in the detection of metal samples in The success of application. But to contain a variety of metals, non-metallic elements of agricultural products, due to its complex composition system, sample preparation, composition distribution, using LIPS for quantitative detection, there still exist some difficulties and challenges. In the laser ablation mechanism of agricultural products on the surface of the metal sample is more complex than. Double atomic radicals CN and C atoms generated by radiation is not clear; and there are serious matrix effect detection process, measurement accuracy obtained by the standard curve of the traditional calibration method and the linear range is poor. In order to explore the application of LIPS technology in the field of agricultural products detection, this paper first laser on graphite induction the plasma spectra and time were studied to distinguish the characteristics to obtain the basic rules of LIPS detection of single carbon materials; then, the combination of LIPS and two linear regression, BP neural network of two set The measurement method, realizes the quantitative of the Chinese herbal medicine angelica root samples are analyzed; and the use of LIPS technology combined with support vector machine algorithm, to achieve a reliable classification and identification of 6 tea samples. The main work and conclusions are as follows: (1) the laser induced plasma graphite CN (B2 + X2 Sigma? Study on characteristics of evolution and C +) atomic radiation spectrum of time. The results showed that CN (B2 + X2 +?) emission duration is 3 times that of C atoms, strength reached a maximum value at the time of 0.2? S to 0.8? S. With the change of the delay time, the electron temperature decreases from 11807 K to 8755 K, the vibrational temperature decreases from 8973 K to 6472 K, the rotational temperature is reduced from 7288 K to 4491 K. (2) to study the effect of laser energy on laser induced CN and C graphite plasma atomic radiation. The results showed that the excitation threshold and spectral characteristics showed different characteristics of CN molecules and C atoms. At low excitation Light energy, electron excitation temperature, CN emission electron density, vibrational and rotational temperatures increased rapidly; at high laser energy, CN emission electron excitation temperature and electron density increased slowly, because the dissociation effect of plasma shielding and CN molecular vibration, temperature and rotational temperature tends to saturation; while the size three kinds of temperature at the same time as the TelecTvibTrot. experimental measurements of the electron density of laser induced plasma graphite, the electron density of 1017? 1018 cm-3. (3) carried out in samples of LIPS detection technology research. Analysis of two lines of characteristic spectral lines from Al elements in the root samples of Angelica the 308.22nm and 309.27nm as two yuan, the establishment of the linear regression equation, regression test results and the ICP-AES value of the good fitting results. This shows that the quantitative two linear regression method based on can To obtain better measurement precision, is suitable for the detection of trace elements in agricultural products in LIPS technology. (4) were selected in the Ca spectra of LIPS samples in angelica root, characteristics of Mg and K elements spectral line Ca II 317.99 nm as the analysis index of spectral intensity of Mg I 517.27 and nm K 769.90 I nm, drawing Ca, BP neural network method for quantitative analysis of Mg and K elements of the calibration curve was established in Chinese angelica root samples of calcium, magnesium and potassium in the new method, the spectral analysis of the content of Ca, R2 Mg and K linear fitting coefficient were respectively 98.96%, 99.45% and 99.11%.. The results show that the analysis based on Ca the quantitative method of BP neural network, Mg and K have higher accuracy and stability, relative error and the relative standard deviation is smaller, verify the feasibility and superiority of the LIPS quantitative analysis using BP neural network method. (5) carried out research on LIPS technology of tea sample detection, combined with support Vector machine algorithm successfully realized the classification of tea samples. The relative intensity discriminant selects 8 spectral lines as the analysis index to build the support vector machine. The discriminant model of the logit model on 300 samples of the training set, the 297 samples of the type of recognition, recognition correct rate of 99%. in addition to 300 samples of test set were identified, 291 samples of which were correct species identification, the correct recognition rate reaches 97%. with LIPS support vector machine algorithm technology has good application prospect in agricultural product identification.

【学位授予单位】：重庆大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：S121

【相似文献】