Elman回归神经网络同时定量测定三种酚类化合物

应用Elman回归神经网络(ERNN)对光谱严重重叠的对-硝基苯酚,邻-硝基苯酚和2,4-二硝基苯酚体系的同时定量测定进行了研究,并与多变量线性回归(MLR)法作了比较。编制了PERNN和PMLR程序执行有关计算。通过最佳化确定了Elman回归网络的结构和参数。ERNN和MLR法所有组分的相对预测标准偏差(RSEP)分别为3.1%和2 027.3%,实验结果显示对于分辨严重重叠光谱本法是成功的。ERNN法是解决局部最小和提高收敛速度的一种有价值的工具,亦可用于分析全光谱而不只限于选取少数特征值。本法为不经预先分离同时测定严重重叠的分子光谱体系提供了新的途径。     

Measurement of chlorophyll in water based on laser-induced fluorescence spectroscopy: Using spiral-wound unclad optical fiber system and partial least squares regression

The feasibility of laser-induced fluorescence (LIF) method for chlorophyll concentration determination in water was investigated in this study. The chlorophyll fluorescence spectra were received by a spiral-wound unclad optical fiber system. Baseline and smoothing correction methods were studied. Statistical equations were established between reference data and fluorescence spectra by linear regression and partial least squares (PLSs) regression. The PLS regression outperformed linear regression by higher correlation coefficient of 0.9996063 and root mean square error value of 0.0221807. Based on the results, it was concluded that laser-induced fluorescence method with a spiral-wound unclad optical fiber system was suitable for chlorophyll concentration determination in water, and PLS regression was a reliable method for fluorescence spectra analysis.     

多元校正分光光度法同时测定兔血清中内的β-内酰胺类抗生素

头孢呋辛钠、头孢曲松钠、头孢噻肟钠和头孢唑啉钠在紫外区均有吸收,且吸收光谱重叠而相互干扰,难以直接进行分析.在pH 2.09的Britton-Robinson(B-R)缓冲溶液中采用紫外分光光度法对其混合溶液进行测定,所得的波谱数据用化学计量学多元校正方法进行了处理.四种β-内酰胺类抗生素的的线性范围分别为1.0～20.0μg·mL-1 ,2.0～20.0μg·mL-1,2.0～20.0μg·mL-1和1.0～18.0μg·mL-1.利用此方法还测定了兔血清中的四种β-内酰胺类抗生素,与高效液相色谱法所获得的结果进行比较,两者无显著性差异.     

Classification of iron‐based inks by means of micro‐Raman spectroscopy and multivariate data analysis

In this work, multivariate data analysis methods were applied to the analysis and interpretation of micro‐Raman spectra, collected from a broad set of historical iron‐based ink samples, previously characterised for the content of organic acids (gallic acid, ellagic acid and protocatechuic acid). The proposed method relies on principal component analysis of the noisy spectra typically obtained on original, degraded, organic samples, where fluorescence could affect the Raman signal. The signal components could be distinguished from the noise components and then used to build a linear discriminant analysis (LDA) model, achieving separation of the spectra into three classes. Selection of pure signal factors also improved effectiveness and performances of partial least square regression (PLS) algorithms, allowing quantification of condensed tannic acid residuals. Application of multivariate methods to discriminate signal from noise removes the need for spectral data manipulation (filtering, smoothing and differentiating). The obtained classification method for discrimination of historic inks and the regression method for determination of condensed tannic acid residuals supports the use of Raman analysis of fluorescing organic materials, and may provide information to scholars on ink composition and potentially on its provenance. Copyright © 2013 John Wiley & Sons, Ltd.     

LIF技术与PLS-DA算法联合辨识矿井涌水水源类型的研究

快速的矿井涌水水源辨识对于矿井的水灾预警及灾后救援意义重大。常规方法使用离子浓度做为判别因子,耗时过长,因此提出一种激光诱导荧光光谱(LIF)技术与偏最小二乘判别分析(PLS-DA)算法联合快速辨识矿井涌水水源类型的方法。实验使用405 nm激光对被测水体进行激发,获取矿井5个不同含水层100组水样的荧光光谱,根据光谱曲线特征,对数据进行压缩处理,获取合适的光谱数据。每种水样使用15组共75组光谱数据作为建模集,剩余的25组水样的光谱数据作为测试集。为验证实验结果,设计了簇类的独立软模式(SIMCA)算法与PLS-DA算法构建的实验模型进行对比。实验发现矿井不同含水层水样的荧光光谱差异较大,在不进行任何预处理的情况下,以PLS模型为基础的PLS-DA算法较SIMCA算法的建模正确率高,达到了100%,其校正及验证结果与实际分类变量的相关系数均大于0.951,校正集均方根误差(RMSECV)和验证集均方根误差(RMSEP)均小于0.123,利用模型对测试集中五种水样样本的识别正确率均为100%。     

POSC-PLS方法及其在电子吸收光谱定量分析中的应用

针对光谱数据局部效应显著,变量个数多,彼此间常存在严重的复共线性,构建了一种基于分段正交信号校正(piecewise OSC)的偏最小二乘(PLS)回归,即POSC-PLS方法。它以近邻分段方式进行逐个波长点的正交信号校正,剔除光谱矩阵中所含的各种噪声信号,将去噪后的光谱矩阵作为新的自变量矩阵,再利用偏最小二乘方法建立校正模型。将该法应用于多环芳香烃电子吸收光谱的多组分定量关系建模,效果良好。所建模型的预报性能优于其他方法,而且模型所需PLS成分数减少,模型更简洁。     

Simultaneous Determination of Mefenamic and Tolfenamic Acids in Real Samples by Terbium-Sensitized Luminescence

A simple luminescent methodology for the simultaneous determination of mefenamic and tolfenamic acids in pharmaceutical preparations and human urine is proposed. Since the native fluorescence of both analytes is not intense, the method takes advantage of the lanthanide-sensitized luminescence, which provides a higher sensitivity. Due to the strong overlapping between the luminescence spectra of both terbium complexes, the use of luminescence decay curves to resolve mixtures of the analytes is proposed, since these curves are more selective. A factorial design with three levels per factor coupled to a central composite design was selected to obtain a calibration matrix of thirteen standards plus eight blank samples that was processed using a partial least-squares (PLS) analysis. In order to assess the goodness of the proposed method, a prediction set of synthetic samples was analyzed, obtaining recovery percentages between 90 and 104?%. Limits of detection, calculated by means of a new criterion, were 14.85?μg L^?1 and 15.89?μg L^?1 for tolfenamic and mefenamic acids, respectively. The method was tested in a pharmaceutical preparation containing mefenamic acid, obtaining recovery percentages close to 100?%. Finally, the simultaneous determination of both fenamates in human urine samples was successfully carried out by means of a correction of the above-explained model. No extraction method neither prior separation of the analytes were needed.     

辐照处理对油菜籽光谱特性影响的研究

针对油菜籽经过核辐照处理后其光谱反射特性会发生改变的特点,提出了应用可见／近红外光谱技术进行油菜籽的快速无损鉴别。利用偏最小二乘法和BP神经网络建立鉴别模型,并比较了不同光谱预处理方法、主成分数据变换方式及隐含层节点数对模型预测结果的影响。实验采用五种剂量辐照(50, 100, 150, 200Gy和不经核辐照处理)的油菜籽共135个样本进行建模,49个进行预测。结果显示,最优模型是原始光谱数据先经过中值滤波平滑法、附加散射校正及一阶求导法预处理。经PLS方法提取6个主成分经自然对数变换后,选取神经网络隐含层结点数为4个或9个。最优模型对是否经过核辐照处理的样本识别率达100%,对核辐照剂量预测精度为85.71%, 说明提出的方法可以用于评估核辐照处理对油菜籽光谱特性产生的明显影响。     

杂交稻种宜香725纯度的可见-近红外反射光谱鉴定

提出了一种基于可见-近红外光谱技术快速、无损鉴定杂交稻种纯度的新方法.以FieldSpec(R)3地物光谱仪采集纯度在90%～99%范围内的杂交稻种(宜香725)光谱数据90份,随机分成校正集(75份)和检验集(15份).根据其在380～2 400 nm的反射光谱,以偏最小二乘算法(PLS)建立了回归模型,并比较了不同光谱预处理方法对模型的影响.分析表明采用一阶导数结合标准归一化处理能最有效地提取光谱信息,此时PLS模型校正集决定系数与检验集决定系数分别为0.988 4与0.922 7,校正标准误差(SEC)与预测标准误差(SEP)分别为0.002 5与0.006 6.将经一阶导数结合标准归一化处理后的光谱进行PCA降维,以前20个主成份(含原始光谱86.09%的特征信息)为输入变量,建立杂交稻种纯度鉴定的BP-ANN模型.分析表明BP-ANN模型校正集决定系数与检验集决定系数分别为0.995 2与0.936 9,SEC与SEP分别为0.001 7与0.006 1,具有比PLS模型更高的精度.结果表明以可见-近红外技术进行杂交稻种纯度的快速、无损鉴定是可行的,且PCA结合BP-ANN是一种优选方法.     

近红外光谱在发射药成分检测中的应用

采用傅里叶变换近红外光谱法测定发射药中外挥发分和内挥发分的含量.本文提出了一种混合算法,该算法将偏最小二乘法(PLS)和人工神经网络(ANN)结合起来,同时利用马氏距离(Mahalanobis)法对异常样品进行剔除.与传统的多元校正算法PLS和主成分回归(PCR)相比,该算法所建模型的预测精度有明显的提高.结果表明,该算法可以满足发射药成分含量的快速分析的需要.     

应用化学计量学技术研究同时定量分析重叠光谱

在八种化学计量学除噪技术比较研究的基础上,研制了小波包变换Elman回归神经网络方法(WPERNN)用于研究重叠光谱的同时定量测定。结合小波包变换和Elman回归神经网络改进除噪质量及回归能力。通过最佳化,选择了小波函数、小波包分解水平和Elman回归神经网络(ERNN)的结构及参数。两个程序PWPERNN和PERNN被设计执行WPERNN和ERNN方法计算。七种化学计量学方法用于比较研究。实验结果显示WPERNN方法是成功的且优于其他六种方法。     

局部建模方法用于烟草样品的近红外光谱定量分析

采用近红外光谱在主成分空间的距离作为样本相似性的判据,建立了一种用于近红外光谱定量分析的局部建模方法。该方法首先对校正集的光谱进行主成分分析(PCA),然后基于主成分空间中预测样本与校正集样本的距离选择校正子集并建立局部偏最小二乘(PLS)回归模型。对欧氏距离和马氏距离的比较表明,欧氏距离可以更好地表达样本之间的相似性。将所建立的方法用于烟草样品中氯和尼古丁含量的测定,结果表明局部建模方法比常用的全局建模方法具有更好的预测准确性,特别是在低含量成分的预测中具有明显优势。     

应用经验模态分解和独立成分分析解译岩矿光谱

岩矿光谱由多种矿物光谱混合而成,解译岩矿光谱能够得到岩矿的组分信息,且该方法具有快速、方便、不损坏样品的特点。经验模态分解(empirical mode decomposition, EMD)不能直接分离出混合信号中的源信号,独立成分分析(independent component analysis, ICA)要求混合信号数目不小于其所包括的源信号数目。将EMD和ICA两种方法相融合,首先用EMD分解混合信号得到本征模态函数(intrinsic mode function, IMF),再选择一定数目的IMF与混合信号一起组成ICA的输入数据矩阵,经过ICA运算可以获取单一混合信号中的源信号信息,克服了EMD和ICA两种方法各自的缺陷。研究表明,综合应用EMD和ICA方法可以获取单一混合信号中的源信号信息,混合信号中源信号含量越大,得到的源信号近似值越理想。参与ICA分离的IMF数目决定了分离得到的源信号近似值的数目,并且选择的IMF与混合信号相关系数越大,得到的源信号近似值越理想。运用该方法定量分析岩矿光谱,可以获取组成岩矿的矿物信息,比较适用于野外作业岩矿的快速分析鉴定及成分初步分析。     

基于聚乙烯膜包装奶酪成分的NIRS检测研究

考察了聚乙烯包装膜对奶酪光谱的影响,提出了消除包装膜影响的新方案。探讨了近红外反射法直接检测带包装膜的奶酪成分的可行性。通过Norris导数滤波处理光谱,可以基本消除聚乙烯包装膜对奶酪光谱的影响,采用PLS结合MSC、求导等预处理建立了包装奶酪定量分析模型,其脂肪和蛋白质相关系数分别为0.928和0.952;建模标准差分别为0.240和0.355;预测标准差分别为0.326和0.219。与无包装奶酪和未滤波处理包装奶酪的模型比较,结果显示：与无包装奶酪模型差异极小,优于未采用Norris导数滤波处理的包装奶酪模型。实验表明近红外光谱分析技术可以在无损条件下快速检测包装奶酪中脂肪、蛋白质含量。     

Novel sparse component analysis approach to free radical EPR spectra decomposition

Free radicals play important roles in many physiological and pathological pathways in biological systems. These free radicals can be detected and quantified by their EPR spectra. The measured EPR spectra are often mixtures of pure spectra of several different free radicals and other chemicals. Blind source separation can be applied to estimate the pure spectra of interested free radicals. However, since the pure EPR spectra are often not independent of each other, the approach based on independent component analysis (ICA) cannot accurately extract the required spectra. In this paper, a novel sparse component analysis method for blind source separation, which exploits the sparsity of the EPR spectra, is presented to reliably extract the pure source spectra from their mixtures with high accuracy. This method has been applied to the analysis of EPR spectra of superoxide, hydroxyl, and nitric oxide free radicals, for both simulated data and real world ex vivo experiment. Compared to the traditional self-modeling method and our previous ICA-based blind source separation method, the proposed sparse component analysis approach gives much better results and can give perfect separation for mixtures of superoxide spectrum and hydroxyl spectrum in the ideal noise-free case. This method can also be used in other similar applications of quantitative spectroscopy analysis.     

基于拟Monte-Carlo方法测定油类混合物组分含量的研究

汽油、煤油、柴油是由原油加工而成,但其相应馏程特征有所不同,其中汽油的沸程约为35~205 ℃,煤油的沸程约为140~250 ℃,柴油的沸程约为180~370 ℃,同时,其碳链长度有所不同,汽油在C7～C11范围内,煤油在C12～C15范围内,柴油在C15～C18范围内,由于其碳数分布特征有所差异,其荧光光谱也相应有所差别,这是对三类不同油种进行识别与定量检测的基础。由于海洋时常发生油类污染,监测海洋中油类有机物的含量十分重要,采用拟Monte-Carlo方法计算三维荧光光谱特征峰幂次积分,结合最优算法求解最佳特征峰数量及积分区域范围,利用BFGS法解非线性方程组,提出一种光谱重叠的多种矿物油混合物组分含量测定的方法。由于对选定特征区域内确定点列对应峰值的幂次进行累加,对荧光谱线微小变化敏感,从而对组分含量微小变化敏感。同时由于点列的选取,相比于单点测量法,在一定程度上减小了随机误差的影响,可以进一步提高测量灵敏度。以煤油、柴油、汽油为研究对象,将单一油种视为整体,不考虑每种油的具体组分,测量单一油种及混合物的三维荧光光谱及等高线图谱,经最优算法选择六个特征峰进行特征峰幂次积分,测定混合油中组分含量,与峰值法,均值法等单点测量方法相比较,测量灵敏度提高约50倍,实现了混合物组分含量的高精度测量,为不需化学分离直接测定光谱重叠混合物组分含量提供了一种实用算法。     

三维荧光光谱结合Zernike图像矩快速鉴别掺伪芝麻油

为了实现对掺伪芝麻油的快速鉴别,应用FS920荧光光谱仪测定样品的三维荧光光谱数据。将三维荧光光谱图视为灰度图,在没有任何预处理的前提下,直接应用Zernike图像矩提取三维光谱灰度图的特征信息,然后采用类平均法对特征信息进行聚类分析,从定性角度实现掺伪芝麻油的鉴别,并解析其组成成分。最后应用广义回归神经网络(GRNN)对掺伪样本的成分进行定量分析。聚类分析能够以很高的辨识率来识别掺伪芝麻油,并能够正确解析其组成成分。定量模型预测了2组掺伪样本中各成分的相对体积,其平均相对误差分别为2.23%,8.00%,9.70%和9.70%。分析结果表明,Zernike矩能够有效提取光谱的特征信息,光谱数据的Zernike矩特征结合聚类分析以及GRNN模型能够获得良好的定性和定量分析结果,为掺伪芝麻油的鉴别提供了一种新的方法。     

Fixed-point algorithms for constrained ICA and their applications in fMRI data analysis

Constrained independent component analysis (CICA) eliminates the order ambiguity of standard ICA by incorporating prior information into the learning process to sort the components intrinsically. However, the original CICA (OCICA) and its variants depend on a learning rate, which is not easy to be tuned for various applications. To solve this problem, two learning-rate-free CICA algorithms were derived in this paper using the fixed-point learning concept. A complete stability analysis was provided for the proposed methods, which also made a correction to the stability analysis given to OCICA. Variations for adding constraints either to the components or to the associated time courses were derived too. Using synthetic data, the proposed methods yielded a better stability and a better source separation quality in terms of higher signal-to-noise-ratio and smaller performance index than OCICA. For the artificially generated brain activations, the new CICAs demonstrated a better sensitivity/specificity performance than standard univariate general linear model (GLM) and standard ICA. Original CICA showed a similar sensitivity/specificity gain but failed to converge for several times. Using functional magnetic resonance imaging (fMRI) data acquired with a well-characterized sensorimotor task, the proposed CICAs yielded better sensitivity than OCICA, standard ICA and GLM in all the target functional regions in terms of either higher t values or larger suprathreshold cluster extensions using the same significance threshold. In addition, they were more stable than OCICA and standard ICA for analyzing the sensorimotor fMRI data.     

小波多尺度正交校正在近红外牛奶成分测量中的应用

光谱分析中,干扰信号的存在直接影响所建分析模型的质量。基于信号和干扰的不同特性,提出了一种扣除背景和噪声干扰的新方法——小波多尺度正交校正(WMOSC)法。首先将原始光谱进行小波变换(DWT),消除噪声及背景信息,然后采用正交信号校正(OSC)滤除与待测组分浓度无关的全部信息。与单纯的小波变换及正交信号校正相比,WMOSC能有效地扣除背景和噪声干扰,使模型具有更强的抗干扰能力,提高了模型的预测精度。利用该方法对牛奶样品的近红外光谱进行处理,采用偏最小二乘法建立校正模型,其牛奶中脂肪、蛋白质和乳糖的预测均方根误差(RMSEP)分别为0.101 6%,0.087 1%和0.110 7%。实验结果表明该方法能有效地去除干扰,保留有用信息。     

基于拉曼光谱技术的茶树叶片中类胡萝卜素含量的无损快速检测

叶片中类胡萝卜素是植被环境胁迫、光合能力和植被发育阶段的指示器。基于叶片的原位拉曼光谱响应特性对龙井43叶片的类胡萝卜素含量进行了研究,建立了两者之间的定量模型。本文共对315个龙井43叶片样本进行了拉曼光谱采集和分光光度检测。为排除检测过程中受噪声、基线漂移等因素的干扰,运用和比较了五种光谱数据预处理方法提取原始拉曼光谱中与茶叶中类胡萝卜素含量有关的有效信息。基于预处理后的数据建立了偏最小二乘(PLS)回归模型,拉曼光谱与类胡萝卜素含量的建模集和预测集的相关系数(r)分别为0.817和0.786。为进一步研究类胡萝卜素的拉曼光谱响应机理,本文采用连续投影算法(SPA)优选了17个拉曼特征波数建立相应的特征波数模型,模型的建模集和预测集的相关系数(r)分别为0.808和0.777。根据已建立的模型,探究了茶树四个不同叶位的叶片类胡萝卜素含量的变化。发现茶树叶片随着叶龄的增加,类胡萝卜素浓度呈先增后减的趋势。以第2位叶的类胡萝卜素含量最高。进一步验证了模型的可行性以及探索了将该模型应用于茶树叶片树龄和叶位探测的可能性。采用拉曼光谱技术可以实现茶树叶片中类胡萝卜素含量的原位、无损、定量检测。