支持向量回归建立成品汽油通用近红外校正模型的研究

针对目前采用偏最小二乘法建立成品汽油分析模型存在的问题,采用近几年新兴的支持向量回归方法建立了多种汽油标号通用的校正模型,其预测能力优于对应的偏最小二乘法,对汽油研究法辛烷值、烯烃和芳烃的预测标准偏差分别为0.37、1.28%和1.38%,可应用于实际的汽油管道自动调合近红外光谱在线分析.     

基于LVs-LS-SVM近红外光谱检测柴油运动粘度的方法研究

为了研究不同光谱输入方法对柴油运动粘度预测模型的影响,本文对预处理后的柴油光谱进行主成分分析(PCA)得到的前6个主成分(PCs)、建立偏最小二乘(PLS)回归曲线选取有效波长(EWs)和运用偏最小二乘回归(PLSR)建模得到的14个潜变量(LVs),将三者分别输入至最小二乘支持向量机(LS-SVM),对结果进行比较分析表明:LVs-LS-SVM建模得到的结果为柴油运动粘度(R_(Pre)~2=0.839,RMSEP=0.317,RPD=1.54),优于EWs-LS-SVM和PCs-LS-SVM模型,为柴油参数测定便携式仪器的开发奠定基础。     

基于最小二乘支持向量机的色谱指纹图谱预测银杏叶总抗氧化活性

在色谱图基线校正和色谱峰匹配基础上,提出以40个银杏叶提取物HPLC指纹图谱的色谱图轮廓作为输入,相应的提取物总抗氧化活性作为输出,建立最小二乘支持向量机回归模型,并对包含10个样本的测试集进行了预测.最小二乘支持向量机的测试集预测误差均方根(RMSEP)为0.0230,预测结果优于目前普遍使用的误差反向传播神经网络和偏最小二乘回归.与采用色谱峰面积为分析变量的模型预测结果比较表明:采用消除干扰后的色谱图全谱轮廓保留了样本的全部信息,预测结果更好     

粒子群最小二乘支持向量机结合偏最小二乘法用于芝麻油质量的鉴别

结合粒子群最小二乘支持向量机(PSO-LSSVM)与偏最小二乘法(PLS)提出一种基于气相色谱技术的新方法,对芝麻油进行真伪鉴别,并对掺伪品中掺假比例进行定量分析。采用主成分分析法(PCA)对857个样本的脂肪酸色谱数据进行分析,优选主成分作为最小二乘支持向量机(LSSVM)的输入向量。利用粒子群算法(PSO)优化LSSVM,构建芝麻油掺伪鉴别的两级分类模型,同时运用PLS建立掺伪芝麻油中掺伪油脂的定量校正模型,两级分类模型的准确率分别达到了100%和98.7%,定量分析模型的平均预测标准偏差(RMSEP)为3.91%。结果表明,本方法的鉴别准确性和模型泛化能力均优于经典的BP神经网络和支持向量机(SVM),可用于食用油脂加工和流通环节的质量控制,为食用油质量的准确鉴定提供了一条有效途径。     

最小二乘支持向量机方法用于提高低水平量子化学方法计算吸收能的精度

运用B3LYP/STO-3G和ZINDO两种低水平的量子化学方法计算了160个有机分子的UV-Vis吸收光谱, 然后提取合适的物理参数, 并以实验值为基础, 引入最小二乘支持向量机方法以提高吸收能的计算值精度. 结果表明, 最小二乘支持向量机方法可有效提高量子化学计算精度, 体系的吸收能误差均方根分别从0.95和0.46 eV降低到0.16和0.15 eV. 最小二乘支持向量机校正方法的引入可在较少的机时和计算资源下得到比单一的量子化学计算方法更为稳定和精确的计算结果, 且可在现有计算条件下预测现有计算能力达不到的精度. 因此, 将最小二乘支持向量机方法用于量子化学数据分析, 为化学研究准确、 快捷地预测分子性质提供了一种新的研究手段.     

粒子群算法结合支持向量机回归法用于近红外光谱建模

研究了最小二乘法支持向量机(LSSVM)应用于烟丝样品和小麦样品的近红外光谱建模,采用粒子群优化算法(PSO)优化LSSVM的参数。通过对烟草样品和小麦样品的近红外光谱建模和预测,并与常规的偏最小二乘法(PLS)比较发现,PSO-LSSVM法具有更好的预测效果和稳健性。     

基于机器学习方法的丙型肝炎病毒聚合酶NS5B非核苷抑制剂的定量构效关系研究

对89 个苯并异噻唑和苯并噻嗪类丙型肝炎病毒(HCV) NS5B聚合酶非核苷抑制剂进行了定量构效关系(QSAR)研究. 采用遗传算法组合偏最小二乘(GA-PLS)和线性逐步回归分析(LSRA)两种特征选择方法选择最优描述符子集, 然后建立多元线性回归和偏最小二乘线性回归模型. 并首次尝试使用遗传算法耦合支持向量机方法(GA-SVM)对两种特征选择方法所选的描述符子集分别建立非线性支持向量机回归模型. 三种机器学习方法所建模型均得到比较满意的预测效果. 采用LSRA所选的6 个描述符建立的三个QSAR模型对于测试集的相关系数为0.958-0.962, GA-SVM法给出最好的预测精度(0.962). 采用GA-PLS所选的7个描述符建立的三个QSAR模型对于测试集的相关系数为0.918-0.960, 偏最小二乘回归模型的结果最好(0.960). 本工作提供了一种有效的方法来预测丙型肝炎病毒抑制剂的生物活性, 该方法也可以扩展到其他类似的定量构效关系研究领域.     

最小二乘支持向量机结合红外光谱法测定润滑油酸值

在润滑油酸值进行红外光谱法测定中,采用最小二乘支持向量机(LS-SVM)建立了酸值的定量预测模型。用Kennard-Stone方法将30个样本划分为训练集(24个样本)和验证集(6个样本),进行定量预测,并与偏最小二乘法和径向基函数神经网络所建模型的预测进行比较。结果表明:LS-SVM所建模型的预测标准偏差(SEP)最小,仅为0.002;预测值的相对误差为1.3%~5.3%。由此认为LS-SVM所建模型的训练和预测结果均优于其余两种方法所建模型。对5个未知样品的分析结果表明:LS-SVM模型的预测值与化学法实测值之间的相对误差(1.2%~3.1%)也较少。     

气相色谱结合化学计量学用于6种食用植物油的分类

运用气相色谱法对6类植物油(大豆油、花生油、茶籽油、菜籽油、玉米油、橄榄油)的脂肪酸组成进行分析,构建植物油的指纹图谱,对植物油进行鉴别和分类。本工作采用遗传-偏最小二乘法(GA-PLS)筛选出7个有效特征变量作为输入变量,采用主成分分析法(PCA)和有监督模式识别法(径向基函数神经网络(RNF-ANN),线性判别分析(LDA)和最小二乘-支持向量机(LSSVM))进行建模分析。结果表明,PCA能够较好地区分六类植物油,而在植物油种类判别分析中,LDA的预报结果最佳。本文提出的方法能够准确直观地区分植物油种类,可用于食用植物油的鉴别和掺杂食用植物油的鉴定。     

基于离散傅立叶变换的支持向量机光谱定量分析法

研究了火电厂电煤煤粉的近红外光谱特征,提取了前3个主成分和前6个离散傅立叶变换(DFT)系数,结合主成分得分、马氏距离和偏最小二乘(PLS)交互验证方法剔除异常样本,并建立偏最小二乘回归(PLSR)、栅格支持向量机回归(G-SVR)、遗传算法支持向量机回归(GA-SVR)和粒子群算法支持向量机回归(PSO-SVR)等定量分析模型。结果表明,利用DFT系数作为PSO-SVR模型的输入变量,当其进化代数为300,种群规模为20,模型参数c1、c2为1.5,1.7时,性能最优,其中校正集相关系数(RC)为0.990,测试集相关系数(RP)为0.954,定标标准差(SEC)为0.366,测试标准差(SEP)为0.128。该方法准确可靠,已成功应用于近红外在线电煤发热量监测系统,并可推广用于其它较为复杂的近红外在线分析系统。     

汽油族组成的近红外光谱快速测定

以荧光指示剂吸附色谱法（FIA）测定汽油族组成结果为基础，采用近红外光谱和化学计量学方法建立了快速、准确测定催化裂化馏出口汽油族组成（饱和烃、烯烃和芳烃）的分析模型；试验表明，该法分析速度快、重复性好、成本低，特别适用于生产中间控制分析。     

A sequence-based computational model for the prediction of the solvent accessible surface area for α-helix and β-barrel transmembrane residues

Predicting the solvent accessible surface area (ASA) of transmembrane (TM) residues is of great importance for experimental researchers to elucidate diverse physiological processes. TM residues fall into two major structural classes (α-helix membrane protein and β-barrel membrane protein). The reported solvent ASA prediction models were developed for these two types of TM residues respectively. However, this prevents the general use of these methods because one cannot determine which model is suitable for a given TM residue without information of its type. To conquer this limitation, we developed a new computational model that can be used for predicting the ASA of both TM α-helix and β-barrel residues. The model was developed from 78 α-helix membrane protein chains and 24 β-barrel membrane protein. Its prediction ability was evaluated by cross validation method and its prediction result on an independent test set of 20 membrane protein chains. The results show that our model performs well for both types of TM residues and outperforms other prediction model which was developed for the specific type of TM residues. The prediction results also proved that the random forest model incorporating conservation score is an effective sequence-based computational approach for predicting the solvent ASA of TM residues.     

ICA方法与NIR技术用于药片中活性成分含量的测定

用独立分量分析(ICA)方法提取药片近红外光谱数据矩阵的独立成分和相应的混合矩阵, 再用BP神经网络对混合矩阵和药片中活性成分的浓度矩阵进行建模, 提出了新的药片活性成分含量测定的基于独立分量分析-神经网络回归(ICA-NNR)的近红外光谱分析方法. 通过分析独立分量数和网络中间隐层的神经元数对模型性能的影响, 分别建立三类药片定量分析的最优模型. 该方法用于实测的三类药片中活性成分含量的测定, 测试样品集的化学检测值与近红外预测值的相关系数分别达到0.962, 0.980及0.979. 结果表明, 基于ICA-NNR的近红外光谱分析方法对制药业的药片进行定量分析是可行的.     

Simultaneous qualitative and quantitative analyses of the major constituents in the rhizome of Ligusticum Chuanxiong using HPLC-DAD-MS

An HPLC-DAD-MS method was developed for the qualitative and quantitative analysis of the major constituents in Chuanxiong (the dried rhizome of Ligusticum chuanxiong Hort). Twenty compounds including phenolic constituents, alkylphthalides and phthalide dimers were identified using online ESI-MS and comparisons with literature data and standard compounds, and six of them were quantified by HPLC-DAD simultaneously. A comprehensive validation of the method including sensitivity, linearity, repeatability and recovery was conducted. The linear regressions were acquired with R(2) > 0.99 and limit of detection (LOD, S/N = 3) values were between 1.5 and 2.5 ng. The repeatability was evaluated by intra- and inter-day assays, and relative standard deviation (RSD) values were reported within 1.87%. The recovery studies for the quantified compounds were observed in the range of 96.36-102.37% with RSD values less than 2.63%. These phenolic constituents and alkylphthalides, the major constituents in Chuanxiong, are generally regarded as the index for the quality assessment of this herb. The overall procedure is accurate and reproducible, which is considered suitable for the qualitative and quantitative analyses of a large number of Chuanxiong samples.     

硅烷化GC-MS指纹图谱在卷烟品牌分析中的应用

采用硅烷化衍生化法结合气相色谱-质谱(GC-MS)法对卷烟烟丝中的主要化学成分进行检测,获得了21个卷烟样品的烟丝硅烷化GC-MS指纹图谱数据,并应用聚类分析和主成分分析法对烟丝硅烷化GC-MS指纹图谱数据进行综合评价。结果表明,该方法可用于不同品牌卷烟的比较和区分,硅烷化成分的含量分布特征能反映不同品牌卷烟的特性,可为卷烟品牌的风格表征、品质维护和真伪鉴别提供参考。     

Application of gas chromatography/pattern recognition techniques to the problem of identifying Africanized honeybees

Capillary column gas chromatography and pattern recognition techniques were used to develop a potential method for distinguishing heavily and moderately Africanized honeybees from European honeybees on the basis of differences in chemical constitution. The test data consisted of 232 chromatograms of hydrocarbon extracts obtained from the cuticle, exocrine, and wax glands of heavily Africanized honeybees, moderately Africanized honeybees, and European honeybees. Each chromatogram contained 65 peaks corresponding to a set of standardized retention time windows. A linear discriminant function, developed by a nonparametric pattern recognition procedure called the linear learning machine, correctly assigned the chromatograms to their respective classes: European or Africanized. The discriminant was then tested, using an external prediction set of 251 chromatograms. Two hundred forty-seven out of two hundred fifty-one prediction set samples were correctly classified. Since many samples in the prediction set were hybrids, these results clearly demonstrate that information derived solely from the nonisoprenoid hydrocarbons could correctly categorize the bee specimens as to race.     

Lipid classes of in vitro cultivar of Pakistani citrus

The in vitro cultivar of citrus contained 18.0% lipids after 12 weeks of germination of seed. The lipid was analyzed by GC procedure for its fatty acid composition. The oil contained seven major fatty acid constituents which were later identified by GC. The oil was further analyzed for its classes by means of thin layer chromatography and gas chromatography. The major portion of the lipid classes comprised of neutral lipids (93.9%) and polar lipids (6.1%). The identified neutral lipids comprised of hydrocarbon (1.4%), wax esters (1.5%), sterol esters (5.2%), triglycerides (52.3%), free fatty acids (1.3%), 1,3-diglycerides (6.0%), 1,2-diglycerides (5.0%), glycol (15.2%), sterols (6.0%), 2-monoglycerides (6.4%), 1-monoglycerides (5.3%) and the identified polar lipids comprised of phosphatidyl ethanolamine (1.8%) phosphatidyl choline (0.9%) lysophosphatidyl ethanolamine (1.8%) and phosphatidyl inositol (1.1%).     

Plasma phospholipid profiling of a mouse model of anxiety disorder by hydrophilic interaction liquid chromatography coupled to high‐resolution mass spectrometry

Glycerophospholipids (PLs), as amphipathic small molecules and the main constituents of biological membranes, play an important role in several cellular processes, even though their accurate identification from complex biological samples remains a challenge. In this paper, we report a fast and comprehensive HILIC‐ESI‐MS method for the analysis of glycerophospholipid classes using high‐resolution mass spectrometry in negative mode. The final method enabled the quantitative analysis of 130 endogenous PL species in mouse plasma. The application of the method developed was to find differences of plasma PL composition in a mouse model of anxiety disorder. In the case of four PL classes and 35 PL species, significant differences were observed comparing low anxiety‐related behavior with high anxiety‐related behavior groups. The most characteristic trend was up‐regulation in both the PL classes and PL species, and decreases were only detected in two phosphatidylcholines among 35 species in mice having elevated anxiety.