基于向量投影的代谢支持向量机乙烯精馏产品质量软测量建模

针对支持向量机(SVM)增量学习过程中易出现计算速度慢、稳定性差的缺陷,提出了一种基于向量投影的代谢支持向量机建模方法.该方法首先运用向量投影算法对训练样本进行预选取来减少样本数量,提高SVM建模速度.然后将新增样本"代谢"原则引入SVM增量学习过程中,以解决因新增样本不断加入而导致训练样本数量"爆炸"的问题.最后将该方法用于乙烯精馏产品质量软测量建模,实验结果表明,与传统SVM和最小二乘支持向量机(LSSVM)相比,向量投影的代谢SVM具有更好的预测结果.     

Toward an intelligent approach for predicting surface tension of binary mixtures containing ionic liquids

Knowledge of the surface tension of ionic liquids (ILs) and their related mixtures is of central importance and enables engineers to efficiently design new processes dealing with these fluids on an industrial scale. It’s obvious that experimental determination of surface tension of every conceivable IL and its mixture with other compounds would be a herculean task. Besides, experimental measurements are intrinsically laborious and expensive; therefore, accurate prediction of the property using a reliable technique would be overwhelmingly favorable. To do so, a modeling method based on artificial neural network (ANN) trained by Bayesian regulation back propagation training algorithm (trainbr) has been proposed to predict surface tension of the binary ILs mixtures. A total set of 748 data points of binary surface tension of IL systems within temperature range of 283.1-348.15 K was used to train and test the applied network. The obtained results indicated that the predictive values and experimental data are quite matching, representing reliability of the used ANN model for such purpose. Also, compared with other methods, such as SVM, GA-SVM, GA-LSSVM, CSA-LSSVM, GMDH-PNN and ANN trained with trainlm algorithm the proposed model was better in terms of accuracy.     

LSSVM过程建模中超参数选取的梯度优化算法

基于结构风险最小的最小二乘支持向量机（least squares support vector machine, LSSVM）为标准支持向量机（SVM）的约简;训练简易;性能良好。其模型精度受超参数影响;常规的网络搜索法很难搜得最佳超参数。在快速留一法的基础上;以全样本留一预测误差平方和最小化为目标;导出基于梯度的最优化算法;用以优选为LSSVM超参数;进而构建G-LSSVM模型。以柠檬酸发酵过程为算例对G-LSSVM进行检验;结果表明G-LSSVM的超参数选取耗时少;模型稳定性良好;且拟合和预报性能都优于标准SVM和神经网络。有望适用于机理不明、高度非线性、小样本的化工过程建模。     

A new fragment contribution‐corresponding states method for physicochemical properties prediction of ionic liquids

A new fragment contribution‐corresponding states (FC—CS) method based on the group contribution method and the corresponding states principle is developed to predict critical properties of ionic liquids (ILs). There are 46 fragments specially classified for ILs considering the ionic features of ILs, and the corresponding fragment increments are determined using the experimental density data. The accuracy of the developed method is verified indirectly via predicting density and surface tension of ILs. The results show that the FC—CS method is reasonable with an average absolute relative deviation less than 4%. With the calculated critical properties, corresponding states heat capacity (CSHC) and corresponding states thermal conductivity (CSTC) correlations are proposed to predict heat capacity and thermal conductivity of ILs, respectively. The predicted results agreed well with the experimental data. The proposed FC—CS method and the two corresponding states correlations are important for design, simulation, and analysis of new ionic liquid processes. © 2012 American Institute of Chemical Engineers AIChE J, 59: 1348–1359, 2013     

基于差分进化算法-最小二乘支持向量机的软测量建模

软测量技术是解决工业过程中存在的一类难以在线测量参数估计问题的有效方法,该技术的核心是建立优良的数学模型。支持向量机是基于统计学理论的一种机器学习方法,最小二乘支持向量机是一种扩展的支持向量机,相对于支持向量机具有较快求解速度。最小二乘支持向量机存在着参数选择的问题,针对这个问题,采用差分进化算法进行参数选择。提出基于差分进化算法的最小二乘支持向量机应用于软测量建模,并将其应用于对苯二甲酸中对羧基苯甲醛含量测试的软测量建模中,获得了满意的结果。     

基于变异CPSO算法的LSSVM蒸发过程软测量

在分析混沌粒子群优化算法(CPSO)和最小二乘支持向量机(SVM)理论基础上,以某氧化铝厂蒸发过程为对象,采用带有末位淘汰机制的混沌粒子群优化算法优化支持向量机的参数,建立了基于变异CPSO算法的LS-SVM的氧化铝蒸发过程软测量模型,并与PSO-LSSVM、LSSVM模型比较,研究表明,ICPSO-LSSVM模型预测准确,泛化性能好,且该模型预测结果中相对误差小于5%的样本达到92.5%,最大相对误差仅为8.1%,均方差MSE为0.05153,模型具有较高的精度,其现场实施结果表明基本可以实现出口浓度的实时在线预估。     

Density, viscosity, and surface tension of water+ethanol mixtures from 293 to 323K

Density, viscosity, and surface tension of liquids are important physicochemical properties which affect mass and heat transfer in solutions. The density, viscosity, and surface tension of binary mixture of water+ethanol at 293, 298, 303, 308, 313, 318, and 323 K are reported and compared with the available literature data. The findings of these comparisons show how the measured data are reproducible from different laboratories. The molar volume of water+ ethanol mixtures are also calculated using measured density values. The Jouyban-Acree model was used for mathematical correlation of the data. The relative deviation (RD) was used as an error criterion and the RD values for correlation of density, viscosity, surface tension and molar volume data at investigated temperatures are 0.1±0.1%, 10.4±9.5%, 4.2±3.6%, and 0.3±0.3%, respectively. The corresponding RDs for the predicted properties after training using the experimental data at 298 K are 0.2±0.2%, 14.1±15.8%, 5.4±4.6% and 0.4±0.3%, respectively, for density, viscosity, surface tension, and molar volume data. This study shows that the Jouyban-Acree model can correlate/predict physicochemical properties of the mixtures of solvents at different temperatures with acceptable error in calculation.     

The Development of Intelligent Models for Liquid–Liquid Equilibria (LLE) Phase Behavior of Thiophene/Alkane/Ionic Liquid Ternary System

ABSTRACT

This paper develops three models based on artificial neural network (ANN), support vector machine (SVM) and least square support vector machine (LSSVM) algorithm for phase behavior of thiophene/alkane/ionic liquid ternary system. The shuffled complex evolution (SCE) was employed to acquire the optimal magnitudes of hyper parameters (σ² and γ) which are embedded parts of SVM and LSSVM models, and the trial and error was employed to obtain the optimal numbers of neuron and layers for ANN intelligent model. Gathering and using 618 LLE data, the comparison between the optimized version of applied intelligent models in giving the LLE was also made. The findings are indicative of a prefect agreement between the estimation from intelligent models and the experimental data. The finding also reveals that the performance of SVM in prediction of solubility is somewhat better than other intelligent models (i.e., ANN and SVM) as coefficient determination (R2) and root mean squared error (RMSE) are respectively 0.9961 and 0.0447 for test sets of data. This is likely due to the existence of structural risk minimization principle of SVM which is embodied in SVM algorithm and effectively minimizes upper bound of the generalization error, rather than minimizing the training error.     

基于分阶段的LSSVM发酵过程建模

发酵过程建模是研究微生物发酵的重要课题,基于模型可实现被测参量的软测量、系统的优化控制。鉴于引入混合核函数的最小二乘支持向量机在过程建模中具有优良表现,采用基于混合核函数的最小二乘支持向量机建模。但由于发酵过程周期较长,最小二乘支持向量机的全局模型预测精度难以保证,算法复杂度很高,因此提出一种分阶段建模方法。首先,选择表征阶段特性的辅助变量,利用模糊C均值聚类算法对样本数据聚类,将发酵过程分成不同的阶段,然后为各个阶段分别建立最优混合核最小二乘支持向量机局部模型,最后将局部模型合成构成过程的完整模型。将此方法应用于青霉素发酵过程和重组大肠杆菌发酵过程中,验证了该方法的有效性。     

基于LM/SVM方法的二次反应清洁汽油辛烷值预测

提出了一种莱文伯格—马夸特(LM)算法和支持向量机(SVM)有机结合的LM/SVM新算法,并将其应用于基于集总模型的二次反应清洁汽油研究法辛烷值的预测。借鉴复杂反应动力学研究中的集总方法,将汽油研究法辛烷值看成汽油饱和烃集总、烯烃集总、芳烃集总的函数,并采用支持向量机表达该函数。针对支持向量机参数及核函数参数难以选择的问题,通过莱文伯格-马夸特算法搜索支持向量机中的参数,并采取把训练集分割成工作样本和检验样本的策略,从而解决了过拟合的问题。利用经典测试函数对LM/SVM算法的性能测试结果表明:LM/SVM算法不但精度优于文献报道的遗传算法与支持向量机相结合的GA/SVM方法,而且其效率也远高于GA/SVM方法。LM/SVM方法对二次反应清洁汽油研究法辛烷值预测的相对误差绝对值的平均值为0.71%。     

基于最小二乘支持向量机和遗传算法的热式油水两相流含油率建模

利用铂电阻基于流体传热方程测量油水两相流的含油率。为了提高油水两相流含油率的测量精度,研究了油水两相流含油率与电热器前后铂电阻的温差和油水总流量之间的关系,提出用最小二乘支持向量机和遗传优化算法对含油率进行建模,将温差和油水总流量作为模型输入量,含油率作为输出量,通过遗传算法优化最小二乘支持向量机的参数,得到了含油率的最优模型,与含油率理论模型相比,基于最小二乘支持向量机和遗传算法的含油率模型,较好的实现了含油率测量,含油率在5%~60%范围内,含油率的平均测量误差为0.96%。     

STUDIES ON THERMAL BEHAVIOUR OF TETRAALKOXYSILANES

Tetraalkoxysilanes (Compounds A to O) were prepared by reacting SiCI4 with corresponding alcohols in the liquid phase in the presence of pyridine as acid acceptor in benzene solvent. Based on the experimental evidences and theoretical calculations tentative structures had been proposed. Properties such as density, refractive index, viscosity, surface tension, molecular weight, thermal expansion, specific heat, thermal conductivity, decomposition temperature and oxidation stability were determined. Prandtl number, Reynold's number and heat transfer coefficient for tetraalkoxysilanes were calculated and were compared with those of coolanols of Monsanto Chemicals, USA. Thermal stability of the compounds was studied by heating them at 200°C for 100 hours in the presence and in the absence of phenyl-a-naphthylamine and diphenylamine as stabilizers and redetermining some of the properties like density, refractive index, surface tension, viscosity and molecular weight

The variation of the properties due to heat treatment is less significant when they are heated in the presence of a stabilizer. The low variation of physical constants by heat treatment particularly in the presence of a stabilizer suggests that these compounds are thermally stable and can be used in heat transfer and other thermal applications with stabilizers     

三元可燃性液体水溶液闪点的理论预测

为了预测三元互溶可燃性液体水溶液的闪点,分析确定了影响混合溶液闪点的主要因素,如沸点、相对分子质量、相对密度、饱和蒸汽压、表面张力等。将这些因素作为输入变量,应用支持向量机算法对混合溶液闪点与其对应理化参数之间的内在定量相关性进行了研究,建立了三元互溶可燃性液体水溶液闪点的理论预测模型。对预测模型进行了验证,讨论了模型的有效性和可靠性;解释了模型反映的机制,明确了混合溶液闪点的主要影响因素及其重要程度。     

Physicochemical properties of amide–AlCl_3 based ionic liquid analogues and their mixtures with copper salt

The physicochemical properties of three different amide–AlCl_3 based ionic liquid(IL) analogues and their mixtures with copper salt, such as conductivity, viscosity, density and isobutane solubility were determined over a wide range of temperatures.The effects of amide structure, amide/AlCl_3 molar ratio and CuCl modification on these physicochemical properties were investigated.Results showed that the conductivity of amide–AlCl_3 based IL analogues was much lower than that of traditional Et_3NHCl–AlCl_3 IL with same ligand/AlCl_3 molar ratio due to incomplete splitting of AlCl_3, whereas the density and viscosity of amide–AlCl_3 based IL analogues were slightly higher.The viscosity of amide–AlCl_3 based IL analogues was closely related to the amide structure,and followed the order of DMA–AlCl_3AA–AlCl_3NMA–AlCl_3 with same amide/AlCl_3 molar ratio.Meanwhile,the density of amide–AlCl_3 based IL analogues ranked in the following order: AA–AlCl_3NMA–AlCl_3DMA–AlCl_3.Increasing the amide/AlCl_3 molar ratio decreased the conductivity and density, while increased the viscosity.The solubility experiment indicated that the isobutane solubility in NMA–AlCl_3 was highest than that in two other IL analogues.Under the modification of CuCl, the conductivity, viscosity and density of these IL analogues increased, whereas the isobutane solubility decreased.These results provide the foundation for the development of a suitable IL analogue catalyst for isobutane alkylation.     

Accurate Prediction of the Surface Tensions of Ionic Liquids–Amines–Water Mixtures Regarding CO2 Recovery and Sequestration Processes

A support vector machine (SVM) model is presented concerning accurate prediction of the surface tension of the complex mixtures of ionic liquids (ILs)–amines–water using the most important types of amines employed in the gas sweetening plants, such as monoethanolamine, diethanolamine, and methyldiethanolamine. Different ILs, such as [bmim] [BF₄], [bmim] [DCA], [mmim] [dmp], [emim] [dep], [bmim] [BF₄], [bmim] [Br], [bmim] [BF₄], and [bmim] [Br], are considered in the presented model. The effects of most important influencing parameters, such as temperature and mass fraction of amines and ILs, on the surface tension of the aqueous mixtures of ILs and amines are well represented. A comparison of the presented model and the most important models reveals that the suggested SVM model has better performance in terms of accuracy and generalizability. The percentage average absolute deviation between the predicted and experimental surface tensions is about 0.44% suggesting the superiority of the presented model over wide ranges of temperatures and species concentrations.     

电站锅炉省煤器受热面的积灰监测

为了实现电站锅炉省煤器受热面的污染情况在线监测,采用支持向量机算法对省煤器的清洁吸热量进行预测,使用灰狼算法（GWO）和遗传算法对模型进行参数寻优并进行预测精确度的对比。再由清洁吸热量计算清洁因子,根据清洁因子的变化来判断省煤器的积灰状态。以某660MW机组为例,将短吹后的数据作为清洁数据样本进行训练和验证,结果表明灰狼算法比遗传算法预测精度更高,所需训练时间更短。最后利用上述模型来预测长吹前省煤器的清洁吸热量,绘制出清洁曲线图。该模型能较好表现省煤器的积灰情况,为受热面积灰在线监测提供依据。     

一种新的近红外光谱信息区间选择方法

基于间隔策略的信息波长选择是近红外光谱分析中广泛应用的一种方法。针对传统算法忽略非线性因素的缺点,首次考虑将最小二乘支持向量机（least-squares support vector machine,LSSVM）方法应用于间隔选择策略,进而提出了一种新的波长选择方法iLSSVM（interval LSSVM）。该算法克服了传统间隔选择算法依赖于线性模型的缺陷,对存在较强非线性的光谱数据能够准确地选择最优信息区间,极大地减少建模变量并显著改善模型预测精度。应用两组业界标准的光谱数据来验证该算法的性能,并和传统方法进行了比较。实验结果表明,在两组数据集上该算法取得的标准预测偏差（root mean square error of prediction,RMSEP）分别比全谱PLS建模降低了20%和4%,比传统的间隔偏最小二乘算法（interval partial least-squares,iPLS）降低了28%和2%。     

Physicochemical properties of water-alcohol mixtures of a homological series of lower aliphatic alcohols

The results of a study of physicochemical properties (density, viscosity, refractive index, contraction, and surface tension) of water-alcohol (water-ethanol, water-propanol, and water-isopropanol) and alcohol-alcohol (propanol-butanol) mixtures are presented. The ratios of water and alcohol corresponding to the maxima of the viscosity and contraction of the mixtures are determined. A nonlinear character of dependences of the density and refractive index for water-alcohol mixtures versus their composition is demonstrated.     

Volumetric properties of ionic liquids from cubic equation of state: Application to pure and mixture

In this work, we developed a cubic equation of state (CEOS) for modeling the volumetric properties of various ionic liquids (ILs). Two temperature-dependent parameters presented in the CEOS, have been determined from two sets of corresponding states correlations which are based on the critical point data or the surface tension of ILs. The predicted densities of pure ILs were compared with the experimental data over a broad pressure range from 1 to 100 MPa. The total average absolute deviations (AADs) of the calculated densities of 948 data points from the experimental data are 1.82% using the critical property and 0.96% using the surface tension and liquid density as scaling parameters. Furthermore, the proposed CEOS was successfully extended to mixtures including IL + IL and IL + solvent systems. 1282 literature data points for mixtures were taken to check the reliability of the mixture version of the proposed CEOS. The AAD of the calculated densities of the mixtures using the surface tension and liquid density as scaling parameter is 0.37%. Furthermore, the excess molar volumes (V^E) of the studied binary mixtures have been successfully computed by the use of the proposed CEOS.