GC × GC/TOFMS analysis concerning the identification of organic compounds extracted from the aqueous phase of sugarcane straw fast pyrolysis oil

GC × GC/TOFMS was applied in the characterization of organic extracts, which were obtained by solid phase extraction (SPE) with C18 commercial cartridges and liquid–liquid extraction (LLE) using three solvents of increasing polarities (hexane, dichloromethane and chloroform), from the aqueous phase generated during the pyrolysis of sugarcane straw. Many oxygenated compounds were identified and classified as acids, phenols, aldehydes and ketones, with a high predominance of phenolic compounds, mainly catechol. GC × GC/TOFMS coupled to some computational tools allowed the identification and classification of the compounds according to their chemical class, number of carbons, number of substituents in the alkyl chains, molecular weight and main peaks in fragmentation spectra. Some identifications were only possible to detect using spectral deconvolution of the mass spectra, and it was possible to clearly classify a series of alkyl phenols (from C1 to C4 alkyl substitutes) and alkyl benzenediols (probably catechols, from C1 to C3 alkyl substitutes). A sequential combination of the pre-concentration of the analytescombined with high-resolution techniques of separation and identification allowed the total characterization of these compounds in the aqueous phase.     

基于紫外吸收光谱的水质硝酸盐含量LLE-SVR建模

基于紫外吸收光谱的总氮测定方法在分析精度上严重依赖于所建数学模型的预测精度.针对紫外光谱数据波段多,容易引入未知物质干扰光谱信息的问题,提出了局部线性嵌入( local linear embedded,LLE)和偏最小二乘支持向量回归机(least square support vector regression machine,LS-SVR)相结合的紫外吸收光谱数据建模方法.首先利用LLE算法将高维的紫外吸收光谱映射到低维的流形空间,实现高维非线性光谱数据结构的特征提取,并利用LS-SVR建立硝酸盐含量的非线性回归模型.仿真结果表明:利用LLE-SVR方法获得的硝酸盐含量预测模型,其训练样本的相对误差为0.001 9,测试样本相对误差为0.035 8,小于单纯的LS-SVR模型的0.023 3和0.0602.     

Phase equilibria (LLE and VLE) of refining operations for enzymatic biodiesel production via quantum mechanical COSMO‐RS method

Phase equilibria, both liquid–liquid equilibria (LLE) and vapor‐liquid equilibria (VLE), are equally important in the refining operations of biodiesel production. Equilibrium distributions of components at the down‐processing of enzymatic ethanolysis reaction were investigated through quantum mechanical COnductor‐like Screening MOdel‐Real Solvent method with implemented vapor pressure equations. Due to the composite nature of fatty acid ethyl ester (FAEE) components as the biodiesel fuel, multicomponent predictive phase equilibria approach was applied for a model reaction system of refined soybean oil with 30% molar excess amount of ethanol. Simulation results for the LLE at 303 K were in significant agreement with reported experimental data. Minimum boiling azeotropes were found for FAEE‐glycerol binaries at isothermal (at 423 and 473K) and isobaric (at reduced pressures) VLE simulations. Similar results were also found at the VLE of glycerol purification operations for multicomponent real case simulations. The feasibilities of simple and rigorous unit operations were also pointed up in this study. © 2012 American Institute of Chemical Engineers AIChE J, 2012     

基于FIA-ISE联用技术的锅炉水总碱度测量的LLE-SVR建模

工业锅炉水总碱度测量常用指示剂法测定,该方法费时,难于自动化。利用流动注射分析（FIA）技术的快速、准确等优点可以实现对工业锅炉水总碱度的快速测定,将流动注射分析与离子选择性电极联用（FIA-ISE）滴定酸碱溶液可以获取连续变化的电压信号。由于锅炉水中常存在未知物质的干扰,基于局部线性嵌入（Local Linear Embedded,LLE）和偏最小二乘支持向量回归机（Least Square Support Vector Regression Machine,LS-SVR）相结合的电压数据建模方法可以消除未知物质的干扰,同时可以降低电压数据的维数。LLE算法将高维电压信号映射到低维流形空间,实现高维非线性电压数据结构的特征提取,LS-SVR建模方法可以建立总碱度的对数和降维后电压信号之间的非线性回归模型。实际待测数据的计算结果表明:基于LLE-SVR方法获得的锅炉水总碱度回归模型,通过对16组测试样本的分析,得到测量的平均相对误差:NaOH为0.83%,Na₂CO₃为0.66%;对于LS-SVR回归建模,NaOH为0.98%,Na₂CO₃为0.86%。LLE-SVR的运算负担只有LS-SVR的3.3%且测量精度要高于常规的峰面积和总碱度之间的线性拟合方法。     

保持局部邻域关系的增量Hessian LLE算法

Hessian LLE算法是一种经典的流形学习算法,但该方法是以批处理的方式进行的,当新的数据点加入时,必须重新运行整个算法,计算所有数据点低维嵌入,原来的运算结果被全部丢弃。鉴于此,提出了一种保持局部邻域关系的增量Hessian LLE(LIHLLE)算法,该方法通过保证流形新增样本点在原空间和嵌入空间局部邻域的线性关系不变,用其已有邻域点的低维坐标线性表示新增样本点,来得到新增点的低维嵌入,实现增量学习。在Swiss roll withhole和frey_rawface数据集上的实验表明,该方法简便、有效可行。     

基于局部线性嵌入算法的汉语数字语音识别

语音信号转换到频域后维数较高,流行学习方法可以自主发现高维数据中潜在低维结构的规律性,提出采用流形学习的方法对高维数据降维来进行汉语数字语音识别。采用流形学习中的局部线性嵌入算法提取语音频域上高维数据的低维流形结构特征,再将低维数据输入动态时间规整识别器进行识别。仿真实验结果表明,采用局部线性嵌入算法的汉语数字语音识别相较于常用声学特征MFCC维数要少,识别率提高了1.2%,有效提高了识别速度。     

基于局部线性嵌入极限学习机的人脸识别新方法

针对人脸图片数量多、容易受噪声干扰,致使人脸识别的识别速度慢、准确率低的问题,提出一种基于局部线性嵌入极限学习机的人脸识别方法——LLE-ELM算法。利用局部线性嵌入(LLE)算法对人脸数据提取特征,最大限度保留原数据的特征结构,减少数据量,降低计算复杂;采用极限学习机(ELM)算法对提取特征后的数据进行分类;实现人脸识别,输出识别准确率和时长。通过在ORL数据库、Yale数据库、AR人脸库和CASIA-WEBFACE人脸库上的数值实验表明:与PCA、SVM、CNN算法对比,该算法具有较高的识别准确率和较快的识别速度。     

磷酸三丁酯-水-磷酸体系液液相平衡研究

测定了磷酸三丁酯－水－磷酸三元体系在２５℃和５０℃下的液液相平衡数据．应用Ｐｉｔｉｅｒ电解质溶液理论和Ｇｕｇｇｅｎｈｅｉｍ似晶格非电解质溶液理论,提出了该体系中有机相组分活度系数的计算方法,在关联出计算所需的模型参数的基础上,计算了该体系在２５℃下的活度系数值。     

The LLE and a linear mapping

The locally linear embedding (LLE) is considered an effective algorithm for dimensionality reduction. In this short note, some of its key properties are studied. In particular, we show that: (1) there always exists a linear mapping from the high-dimensional space to the low-dimensional space such that all the constraint conditions in the LLE can be satisfied. The implication of the existence of such a linear mapping is that the LLE cannot guarantee a one-to-one mapping from the high-dimensional space to the low-dimensional space for a given data set; (2) if the LLE is required to globally preserve distance, it must be a PCA mapping; (3) for a given high-dimensional data set, there always exists a local distance-preserving LLE. The above results can bring some new insights into a better understanding of the LLE.