用支持向量机识别毒品的太赫兹吸收光谱

在采用太赫兹时域光谱(THz-TDS)技术对9种常见毒品纯品和3种混合物进行实验研究,并得到它们在0.2～2 THz频率范围的特征吸收光谱的基础上,用支持向量机(SVM)对毒品纯品和混合物的太赫兹吸收光谱进行了识别分类.用归一化预处理后的9种毒品和面粉的太赫兹吸收光谱训练libsvm模型.选用与训练光谱不同时间测得的毒品和混合物的太赫兹吸收光谱作为检测光谱,经过归一化预处理之后分别输入到训练好的libsvm模型中进行识别,识别率达100%.识别结果允分表明,用支持向量机可以实现对不同种类毒品的识别和鉴定,为太赫兹光谱技术用于毒品的检测和识别提供了另一种有效的方法.     

用两种不同测量技术对毒品THz光谱的研究

为了使太赫兹(THz)技术在毒品检测中变得更为实用,运用太赫兹时域光谱(THz-TDS)技术和傅里叶变换红外光谱(FTIR)技术分别研究了甲基苯丙胺(MA)、4,5-亚甲基二氧基苯丙胺(MDA)和3,4-亚甲基二氧基-N-甲氧基苯丙胺(MDMA)3种走私毒品的THz吸收光谱,获得了这些毒品不同谱宽的THz光谱,并对两类...     

隐匿危险品高准确度太赫兹光谱识别方法

爆炸物等危险品的分子振动和转动能级在太赫兹频谱段具有独特的指纹谱特性,且太赫兹波对非极性物质及介电材料有较强的透过性及低能性,因此利用太赫兹光谱可以实现障碍物隐匿复杂环境下的危险品无损探测。目前各种相关材料的太赫兹吸收光谱标准库并不完善,且市面上各类太赫兹光谱仪硬件参数不同、检测标准不统一,导致单纯依赖特征吸收峰的识别方法并不可靠。针对上述问题,提出一种不依赖于吸收峰准确性的物质识别技术路线:提取物质在不同频率分辨率、不同障碍物隐匿情况下的太赫兹吸收谱,利用Marr小波变换在频域上展开得到具有特征唯一性的小波频域尺度图,建立样本集;其次,结合迁移学习方法,利用Xception网络对样本集进行训练识别。实验结果表明,此方法可以很好地对不同障碍物隐匿环境中的危险品进行分类识别,识别准确率可达94%。说明此方法的识别准确性不受系统频率分辨率即吸收谱精确度等系统因素影响,为邮件及快递包裹等障碍物隐匿危险品无损检测、定性识别提供了一种新的技术思路。     

钝感RDX及HMX炸药的太赫兹光谱分析

采用太赫兹时域光谱技术（Terahertz time-domain spectroscopy,THz-TDS）对钝感RDX（黑索金）和HMX（奥克托金）炸药在0～2.5THz频段的太赫兹吸收光谱进行了探测.得到了待测样品的太赫兹吸收光谱,确定了其特征吸收峰的位置并与其他研究机构所测吸收谱进行了对比分析.实验结果表明：利用THz-TDS技术可以对炸药进行检测和识别,同时可以对样品组成的细微变化进行分析和鉴别.为利用THz-TDS技术研究其他爆炸性物质,建立爆炸物的指纹谱库提供了科学依据.     

太赫兹光谱技术在气体检测中的应用

相关环境污染信息在太赫兹波段内具有较强的吸收特性,使得利用太赫兹光谱技术探测大气中的污染物成为可能.概述了国内外太赫兹时域光谱技术在气体检测方面的研究历史和发展现状,介绍了不同的气体分子在太赫兹波段的吸收谱,以及对混合气体分子和同素异形体分子的识别.对太赫兹时域光谱技术在气体检测方面的研究方向进行了展望.     

硫酸镁水合物的太赫兹表征

太赫兹时域光谱技术作为一种新型光谱检测手段,可用于包括水合物在内的多种物质的表征。金属硫酸盐具有多种不同的水合物,在太赫兹波段表现出不同的吸收特性,利用太赫兹光谱技术可以对过渡金属硫酸盐的不同水合物进行有效表征。本文利用太赫兹时域光谱系统对硫酸镁及其水合物进行表征。通过对七水硫酸镁进行恒温干燥,得到不同的硫酸镁热解产物并对其太赫兹吸收光谱进行研究。实验结果表明不同水合物表现出差异明显的太赫兹吸收特性。这为硫酸镁水合物的多种应用提供了重要参考,也为过渡金属硫酸盐及其水合物的检测提供了简单、快捷、有效的表征手段。     

基于小波变换的爆炸物太赫兹光谱降噪分析

采用太赫兹时域光谱技术(terahenz time-domain spectroscopy,THz-TDs)对钝感RDX(黑索金)和HMX(奥克托金)炸药在0～2.5 THz频段的太赫兹吸收光谱进行了探测.得到了待测样晶的太赫兹吸收光谱,确定了其特征吸收峰的位置,与其他研究机构所测吸收谱进行了对比分析.结果表明,利用太...     

丁基羟基茴香醚的太赫兹时域光谱检测与分析

利用太赫兹时域光谱（THz-TDS）技术对食品添加剂丁基羟基茴香醚（BHA）及其与聚乙烯不同配比的混合物进行了检测,获得了其在0.2-2.6THz波段的吸收谱和折射率谱。实验结果表明,在此波段内该物质有三个明显的吸收峰,分别位于0.64THz,1.14THz和1.81THz,该指纹谱可用于THz波段该物质的检测与识别。此外,利用最小二乘法对混合物太赫兹光谱峰位的吸收系数和质量分数进行线性拟合,结果表明聚乙烯中丁基羟基茴香醚的含量与太赫兹光谱的相关系数大于0.98。最后,用密度泛函理论对单分子丁基羟基茴香醚进行了数值计算,表明太赫兹指纹谱与分子构型和分子内部运动有关系。     

基于机器学习的太赫兹光谱分析与识别

国内对转基因作物的监管非常严格,但是对转基因作物的检测缺乏快速准确的计量方法。太赫兹时域光谱结合机器学习分类算法可以实现对转基因作物快速有效地检测识别。通过太赫兹时域光谱技术提取了2种转基因油菜种子和一种非转基因油菜种子的太赫兹吸收谱,朴素贝叶斯算法、基于朴素贝叶斯的自适应提升算法、主成分分析结合随机森林算法、主成分分析结合支持向量计算法被应用于转基因油菜种子的太赫兹吸收谱的分类识别。通过实验对比,基于朴素贝叶斯的自适应提升算法获得了高达96.6%的检测准确率。该研究为运用太赫兹光谱技术手段开展转基因作物的快速检测提供方法参考。     

维甲酸和叶酸的太赫兹光谱

利用太赫兹时域光谱技术（THz-TDS）在室温下对两种抗肿瘤维生素类药物维甲酸和叶酸在0.2～1.8 THz范围内进行了光谱测量,得到对应的吸收谱和折射率谱,发现叶酸和维甲酸在太赫兹波段都有明显的特征吸收峰,可作为其在太赫兹波段的指纹谱用于药物分子识别。同时,采用Gaussian03软件进行了单分子的密度泛函理论模拟,模拟中采用B3LYP结合高斯型基组6～31 G（d,p）,理论计算与实验结果基本吻合。另外,采用Gaussian-View3.0对叶酸和维甲酸在THz波段的特征吸收峰进行了指认。结果证明:利用太赫兹光谱技术结合密度泛函理论模拟对两种抗肿瘤维生素类药物分子识别及其新药研制具有一定的参考意义。     

基于太赫兹时域光谱技术的农药残留检测方法

太赫兹时域光谱(THz-TDS)技术是近年来涌现出来的崭新的光谱测量新技术.本研究提出了一种基于THz-TDS技术农药残留检测方法,并以灭多威和乙氧氟草醚两种农药作为实验介质证明此方法的可行性.应用太赫兹时域光谱系统测得了这两种农药的时域光谱信号,利用基于菲涅尔公式的数据处理模型得到了它们在THz波段的折射率谱和吸收系数谱.从实验结果可以看出两种农药在0.2~2.0THz范围内存在明显的特征吸收峰,且差别很大.经分析认为这些吸收峰是由分子的集体振动模式以及分子间相互作用引起,它们是农药分子的指纹吸收光谱,可以应用于分子识别中.本研究证明了THz-TDS技术应用于农药残留检测的可行性,表明其在农药残留检测中具有潜在的应用价值.     

磺苄西林、舒他西林、美洛西林、替卡西林的太赫兹指纹谱

以一类常用的抗生素-青霉素类抗生素作为研究对象,选取4 种具有代表性的药品磺苄西林、舒他西林、美洛西林、替卡西林,基于太赫兹时域光谱（THz-TDS）技术,进行实验研究。通过光谱实验及理论分析,获取药品的太赫兹时域光谱,结合傅里叶变换,获得频域光谱及太赫兹吸收系数曲线。结果表明,4 种药品在0.40~1.60 THz 波段存在明显不同的吸收特征。因此,太赫兹光谱技术十分适合检测抗生素这种化学结构有微小不同的药品,并且可以清晰通过吸收峰的位置分辨出抗生素药类的种类。为国家食品药品监督管理提供一种新的可靠的检测技术,且可以以数据库的形式为药品的鉴定提供标准。     

盐酸吗啡的太赫兹光谱和理论仿真分析

采用太赫兹时域光谱(THz-TDS)技术测量了盐酸吗啡原料药、4种辅料、中间体、成药在0.2~2 THz波段的光谱。盐酸吗啡原料药特征吸收峰为0.75 THz,1.17 THz,1.76 THz,原料药到成药的THz光谱发生了改变,中间体和成药的吸收谱有非常高的一致性,说明包衣对成药的THz光谱没有显著影响。采用密度泛函理论(DFT)对盐酸吗啡分子进行理论计算,理论吸收峰为0.84 THz,1.10 THz,1.50 THz,理论光谱与实测光谱有较好的一致性。这为太赫兹技术用于药物的快速鉴别提供了理论和实验基础,在建立太赫兹指纹光谱、盐酸吗啡的快速鉴别方面有良好的应用前景。     

基于太赫兹时域光谱技术的磺胺甲噁唑多晶型现象

磺胺甲噁唑(SMX)作为典型的一类光谱抗生素,经常被用于敏感菌引起的各类感染。由于其晶体内部结构的不同,SMX 将以两种不同的多晶型形式存在。不同晶型常常具有不同的物理化学性质,这些差异可能会对药物的流动性、稳定性以及药效等有较大影响。新兴发展起来的太赫兹光谱技术对于分子间弱的相互作用及大分子的骨架振动、偶极子的旋转和振动跃迁以及晶体中晶格的低频振动吸收都非常敏感,在药物以及生物等领域获得了重要的应用。利用太赫兹时域光谱(THz-TDS)技术在室温下对两种不同多晶型SMX 类药物在0.2~1.5 THz 范围内进行了光谱测量,得到对应的吸收谱,发现其在太赫兹波段都有明显的特征吸收峰,可作为其指纹谱用于该类药物多晶型的识别。不同温度下(95~296 K)获得太赫兹时域波形以及对应的频域吸收谱图,结果发现时域波形以及频域吸收谱均与温度呈现规律变化。该研究结果表明THz-TDS 技术药物与生物领域研究、区分药物分子多晶型现象等方面具有重要的应用。     

二硝基苯甲酸同分异构体的太赫兹与红外光谱特性

利用太赫兹时域光谱系统(THz-TDS)和傅里叶变换红外光谱仪(FTIR)分别研究了2,4-, 2,5-, 3,4-, 3,5-二硝基苯甲酸的吸收谱.实验结果表明, 4种同分异构体的吸收光谱在红外波段(1 400~1 800 cm-1)表现出相似性, 而在太赫兹波段(0.3~2.2 THz)却存在非常明显的区别.利用密度泛函理论(DFT)对4种物质的吸收频谱进行计算, 并根据计算结果对吸收光谱的相似性和差异性进行解释.太赫兹时域光谱技术为鉴别物质的同分异构体提供了一种可行的手段.     

Optical properties of selected Chinese Traditional Patent Medicines and Western Medicines terahertz range

The spectral characteristics of Sang Ju Gan Mao Granules, Gan Mao Jie Du Tablet, Fu Fang Jin Yin Hua Granules and Compound Paracetamol, and Amantadine Hydrochloride Tablets are studied by terahertz time-domain spectroscopy (THz-TDS) technologies in the spectral range of 0.2—1.7 THz. The absorption spectra and the indexes of refraction are obtained. It can be seen that the three samples of Chinese Traditional Patent Medicines have almost the same absorption peaks at 1.44 THz. However, the absorption peaks of the three western medicine samples are at 0.73, 1.01, and 1.18 THz, respectively. The large spectral differences between different drugs are strong evidences that THz time-domain spectroscopy is a useful fingerprint technique in the study of pharmaceutical compounds and the crack down on the fake medicines.     

Terahertz Multivariate Spectral Analysis and Molecular Dynamics Simulations of Three Pyrethroid Pesticides

The terahertz (THz) multivariate spectral characteristics and the molecular dynamics of three pyrethroid pesticides, including deltamethrin, fenvalerate, and beta-cypermethrin, were studied in this paper. THz spectra of the pesticides were measured in frequency range of 0.06–3.5 THz by using THz time-domain spectroscopy (THz-TDS). To improve the THz spectral quality, the wavelet threshold de-noising (WTD) method was used to remove spectral noise and the spectral baseline correction (SBC) method was used to remove baseline drift. Specific absorption peaks were observed in the processed THz spectra of the three pesticides. Deltamethrin showed three peaks at 0.90, 1.49, and 2.32 THz. Fenvalerate had five peaks at 1.13, 1.43, 1.61, 1.98, and 2.58 THz. Beta cypermethrin had four peaks at 1.27, 1.84, 2.12, and 2.92 THz. The density functional theory (DFT) was used to characterize the molecular dynamics and formation mechanism of the absorption peaks. Results showed that there was a good matching effect between the THz experimental spectra and the DFT quantum calculation spectra. Based on the characterized fingerprint absorption peaks, the linear addition model was used to simulate the THz spectra of mixed pesticides. The simulated spectra of multicomponent pesticides were demonstrated to be in good agreement with those obtained by THz-TDS. By analyzing the absorption peaks of THz spectra, the composition and concentration of multicomponent pesticides could be determined. The proposed strategy presented an analytical methodology for studying the THz spectral characteristics of pesticides. In addition, this work provided experimental and theoretical basis for the detection potential of pesticides in agricultural products based on THz technology.     

Monitoring PM2.5 in the Atmosphere by Using Terahertz Time-Domain Spectroscopy

The real-time monitoring of the air pollution with multiple sources is of great significance for pollution control and environmental protection. In this paper, we presented a study of terahertz time-domain spectroscopy (THz-TDS) as a direct tool for monitoring the component and content of PM2.5 in atmosphere. Due to the THz absorption, the intensities of the peaks in THz-TDS decreased with the augment of PM2.5 and were proportional to the PM2.5 content. The ratio of absorbance A to PM2.5 reflected a basically unchanged tendency, indicating the little change of principal elements under the pollution degree. In the high-pollution condition, a lot of SO₂ from vehicle and factory was emitted into air. The elements, such as S and O from anions, had a stronger absorption effect in THz range. Based on the absorbance spectra, the absorption tendencies with PM2.5 over the whole range were validated by principal component analysis and the quantitative model with a high correlation was built by using back propagation artificial neural network. BPANN model improved the precision of linear fitting between peak intensities and PM2.5. The research demonstrates that THz-TDS is a promising tool for fast, direct, and reliable monitoring in environmental applications.     

Experimental and Theoretical Study on Terahertz Absorption Characteristics and Spectral De-noising of Three Plant Growth Regulators

Accurate identification of terahertz (THz) absorption peaks of biological macromolecules is of great significance in practical applications. In this work, the experimental and theoretical fundamentals of detecting three plant growth regulators (PGRs), including 6-Benzylaminopurine (6-BA), paclobutrazol (PBZ), and maleic hydrazide (MH) were investigated by using THz time-domain spectroscopy (THz-TDS). THz absorption coefficient and refractive index in frequencies of 0.06–4 THz were obtained. The wavelet threshold de-noising (WTD) method was used to remove spectral noise and improve the signal-to-noise ratio (SNR). The density functional theory (DFT) was applied to the molecular characterization and theoretical calculation of PGRs. Experimental results showed that the three PGRs had unique characteristic absorption peaks. Based on the sym4 wavelet function and four-layer wavelet decomposition, the de-noising performance of hard threshold WTD was better than that of soft threshold WTD. The spectra processed by hard threshold de-noising achieved higher peak SNR (6-BA: 40.22, PBZ: 37.73, MH: 34.83) and lower root mean square error (6-BA: 0.41, PBZ:0.40, MH:0.54). In addition, the characteristic absorption and anomalous dispersion of 6-BA were found at 2.08 and 3.00 THz, those of PBZ were shown at 0.71, 1.30, 1.88, and 2.67 THz, and those of MH were shown at 2.34 THz. The absorption peaks in THz spectra processed by hard threshold WTD were demonstrated to be in good agreement with the simulation results of DFT. These results show the effectiveness of WTD in THz spectral de-noising and the feasibility of using THz-TDS to detect PGRs.