应用经遗传算法优化的BP神经网络预测催化裂化装置焦炭产率

焦炭是催化裂化装置的主要副产物,准确预测催化裂化焦炭产率对提高装置的操作平稳度和经济效益具有重要意义。人工神经网络(ANN)具有强大的自学习和自适应能力,在非线性预测方面具有明显的优势。本研究将遗传算法(GA)与BP神经网络相结合,基于某炼厂催化裂化装置的生产数据,分别从原料、催化剂和操作条件3个方面选取28个关键影响参数建立了催化裂化焦炭产率预测模型,分别将BP神经网络和经遗传算法优化的BP神经网络(GA-BP)的预测结果与工业数据进行对比。结果表明,经遗传算法优化的预测模型无论在预测结果的准确性还是稳定性方面效果更好。最后,本研究还通过考察原料残炭、反应温度等单一关键参数对焦炭产率的影响,进一步证明了经遗传算法优化的BP神经网络预测模型的准确性。     

Liquid‐Phase Alkylation of Benzene with Propylene Catalysed by HY Zeolites

Benzene alkylation with propylene over a Ca modified HY zeolite to obtain cumene has been studied. Time on stream behavior of the catalyst was quite steady without any sign of deactivation and the results were reproducible. The external mass transfer does not influence the cumene reaction. The pore diffusion has a substantial influence on the cumene reaction only in the final region of the reactor, where the olefin concentration is very low. A simple power law kinetic model fits the experimental data significantly better than other models. A general time and space dependent model was developed to analyze the performance of a two‐phase downflow fixed bed reactor used for alkylation of benzene with zeolite catalyst. The model could be instructive in exploratory simulations evaluating the conditions for benzene alkylation in liquid phase.     

Alkylation of Benzene with Isopropyl Alcohol over SAPO-5 Catalyst in an Integral Pressure Reactor

The alkylation of benzene with isopropyl alcohol was studied in an integral pressure reactor over silicon substituted aluminophosphate molecular sieves, SAPO-5. The influence of various process parameters such as temperature, pressure, time on stream, weight hourly space velocity, and mole ratio of reactants on cumene yield and selectivity were investigated. The activity of SAPO-5 was compared with that of Hbeta for this reaction under similar conditions and in the same reactor. At pressures higher than atmospheric, almost the theoretical maximum yields of cumene were achieved on this SAPO-5. Among the diisopropyl benzenes formed by the alkylation of cumene, the meta-isomer was found to form in a significant amount followed by the para-isomer. The ortho-isomer with relatively high strain energy of 4.26 kcal/mol was almost negligible. The cumene yield goes through a maximum in the temperature range 498-543 K studied. Cumene selectivity was found to decrease at higher temperatures, higher pressures and lower benzene to isopropanol mole ratios.     

On-Line Prediction of a Fixed-Bed Reactor Using K-L Expansion and Neural Networks

An on-line prediction scheme combining the Karhunen-Love expansion and a recurrent neural network for a wall-cooled fixed-bed reactor is presented.Benzene oxidation in a pilotscale,single tube fixed-bed reactor is chosen as a working system and a pseudo-homogeneous twodimensional model is used to generate simulation data to investigate the prediction scheme presentedunder randomly changing operating conditions.The scheme consisting of the K-L expansion andneural network performs satisfactorily for on-line prediction of reaction yield and bed temperatures.     

合成异丙苯固定床鼓泡反应器模拟

建立了分子筛催化剂上合成异丙苯固定床鼓泡反应器的双区数学模型，经实际运转数据验证，说明该模型是适宜的。以此模型考察了进料方式、外循环以及进料苯烯比的影响，给出了反应器内的温度分布，并提出了固定床鼓泡反应器的改进方案。     

Modeling the oxidative coupling of methane using artificial neural network and optimizing of its operational conditions using genetic algorithm

The effect of some operating conditions such as temperature, gas hourly space velocity (GHSV), CH₄/O₂ ratio and diluents gas (mol% N₂) on ethylene production by oxidative coupling of methane (OCM) in a fixed bed reactor at atmospheric pressure was studied over Mn/Na₂WO₄/SiO₂ catalyst. Based on the properties of neural networks, an artificial neural network was used for model developing from experimental data. To prevent network complexity and effective data input to the network, principal component analysis method was used and the number of output parameters was reduced from 4 to 2. A feed-forward back-propagation network was used for simulating the relations between process operating conditions and those aspects of catalytic performance including conversion of methane, C₂ products selectivity, C2 yielding and C₂H₄/C₂H₆ ratio. Levenberg-Marquardt method is presented to train the network. For the first output, an optimum network with 4-9-1 topology and for the second output, an optimum network with 4-6-1 topology was prepared. After simulating the process as well as using ANNs, the operating conditions were optimized and a genetic algorithm based on maximum yield of C₂ was used. The average error in comparing the experimental and simulated values for methane conversion, C₂ products selectivity, yield of C₂ and C₂H₄/C₂H₆ ratio, was estimated as 2.73%, 10.66%, 5.48% and 10.28%, respectively.     

乙炔加氢反应过程混合建模与优化

针对传统单一建模方法所构建的乙炔加氢反应器数学模型存在预测性能无法满足工业实际应用需求的问题,提出了一种机理与神经网络嵌套的建模方法,充分利用机理模型包含的能质约束信息降低神经网络模型的约束违反度,得到了能够良好描述实际工业乙炔加氢反应过程特性的混合模型。基于反应器混合模型,研究了以运行效益为目标函数的优化问题。主要决策变量包括：一段反应器进料中氢气与乙炔的摩尔比（R _H/A）、进料温度和反应器运行周期等几个关键参数。针对反应器长期运行后,催化剂活性降低造成的处理能力下降的问题,提出了反应温度补偿机制和R _H/A并行调节的运行优化策略,并采用序列法对反应器运行周期进行离散化处理。通过引入差异化变异策略、潜在解替代策略对两阶段差分算法进行改进,采用增量式编码法结合改进两阶段差分算法,对优化问题进行求解。结果证实了优化策略与改进算法的有效性,并据此确定了反应器最佳运行方案。     

Transalkylation of diisopropylbenzene with benzene over SAPO‐5 catalyst: a kinetic study

BACKGROUND: The objective of this work was to improve the selective yield of cumene by transalkylating the (diisopropylbenzene) DIPB formed in benzene alkylation with propene and select the optimum process conditions for this transalkylation over SAPO‐5 catalyst. RESULTS: Higher benzene/DIPB mole ratios and higher space velocities were found to give better cumene selectivity. An apparent activation energy value of 130.2 kJ mol^?1 was obtained for the reaction with SAPO‐5 as the catalyst. CONCLUSION: SAPO‐5 shows promise when loaded with a small quantity of platinum (0.005% w/w) in the transalkylation of commercial DIPB (cumene column bottoms) with benzene. The kinetic investigations help in selecting optimum process operating conditions for maximizing the cumene yields in transalkylation of commercial DIPB over SAPO‐5. Copyright © 2008 Society of Chemical Industry     

多级库存集成模型优化预测研究

以供应链多级库存为研究背景,建立以生产为中心的供应、生产和销售3个环节的多级库存集成化动态模型,并对模型进行仿真优化。结合训练神经网络的混合算法GA-BP算法,提出了基于遗传算法与人工神经网络相结合的优化预测模型。最后给出实例说明GA-BP算法优化预测模型的求解过程,验证了模型的可行性。     

基于GA-ANN的干气制乙苯反应器模拟预测

干气制乙苯反应器在生产过程中催化剂不断失活,为了使产品质量满足要求,需要不断调解反应器温度,保证催化剂的活性。本文利用现场生产数据,采用基于遗传算法的人工神经网络对干气制乙苯反应器温度进行模拟预测,仿真结果显示模型准确度高,运行速度快,此项技术应用前景广阔,能够对企业的生产管理带来很好的收益。     

Melt index prediction by RBF neural network optimized with an adaptive new ant colony optimization algorithm

Melt index (MI) is a crucial indicator in determining the product specifications and grades of polypropylene (PP). The prediction of MI, which is important in quality control of the PP polymerization process, is studied in this work. Based on RBF (radial basis function) neural network, a soft‐sensor model (RBF model) of the PP process is developed to infer the MI of PP from a bunch of process variables. Considering that the PP process is too complicated for the RBF neural network with a general set of parameters, a new ant colony optimization (ACO) algorithm, N‐ACO, and its adaptive version, A‐N‐ACO, which aim at continuous optimizing problems are proposed to optimize the structure parameters of the RBF neural network, respectively, and the structure‐best models, N‐ACO‐RBF model and A‐N‐ACO‐RBF model for the MI prediction of propylene polymerization process, are presented then. Based on the data from a real PP production plant, a detailed comparison research among the models is carried out. The research results confirm the prediction accuracy of the models and also prove the effectiveness of proposed N‐ACO and A‐N‐ACO optimization approaches in solving continuous optimizing problem. © 2010 Wiley Periodicals, Inc. J Appl Polym Sci, 2010     

甲醇制芳烃K-means-PSO-SVR局部建模及优化

针对甲醇制芳烃（MTA）过程数据样本趋同、维度高、非线性、强耦合、局部差异大的特性,提出了一种K-means-PSO-SVR的局部建模方法,用以解决单一全局模型预测精度低,鲁棒性不强的问题。该方法首先用K-means算法对样本空间的数据进行聚类,实现对样本空间k个区域的划分,再用经过粒子群优化算法（PSO）优化过超参数的支持向量回归算法（SVR）在划分好的样本空间上建立相互独立的局部模型,最终将建立的k个相互独立的局部模型组合起来组成覆盖整个样本空间的集成模型。在不同噪声水平下将K-means-PSO-SVR方法的建模效果与单一全局SVR、BP神经网络和线性回归3种算法的建模效果进行了比较分析,结果表明：K-means-PSO-SVR局部建模方法的性能在所有水平的噪声下都明显优于其他3种建模方法,并且该方法对噪声具有很强的鲁棒性。以建立的数据模型为基础优化了两段式固定床甲醇制芳烃的关键工艺参数,并用5次独立重复实验验证了优化结果的可靠性,得出当一段温度为446.2℃、二段温度为467.3℃、甲醇体积空速为0.4h^-1、压力为0.64MPa时反应产物中苯、甲苯和二甲苯（BTX）的总收率最高,最高收率为44.30%。     

VPT法制备MCM-22分子筛催化剂颗粒及其烷基化性能

采用汽相法（VPT）制备了以成型干胶颗粒为基体的MCM-22分子筛催化剂.利用XRD、SEM和BET测试技术对所制备催化剂进行了物相表征、形貌观察和织构分析,并以苯和丙烯烷基化合成异丙苯为模型反应,在液-固固定床反应器上对所制备催化剂的性能进行了评价.结果表明：汽相法制备的MCM-22分子筛催化剂不仅具有较好的结晶度和较高的微孔表面积,而且具有更好的晶体生长取向性和更加致密的结构;在烷基化反应中也具有很好的活性、选择性和稳定性,特别是对目的产物异丙苯的选择性明显高于水热法（HTS）合成的催化剂,在异丙苯合成上具有很好的应用前景.     

A novel approach to the design and operation scheduling of heterogeneous catalytic reactors

A number of studies have been conducted to reduce the overall level of catalyst deactivation in heterogeneous catalytic reactors, and improve the performance of reactors, such as yield, conversion or selectivity. The methodology generally includes optimization of the following: (1) operating conditions of the reaction system, such as feed temperature, normal operating temperature, pressure, and composition of feed streams; (2) reactor design parameters, such as dimension of the reactor, side stream distribution along the axis of the reactor beds, the mixing ratio of inert catalyst at each bed; and (3) catalyst design parameters, such as the pore size distribution across the pellet, active material distribution, size and shape of the catalyst, etc. Few studies have examined optimization of the overall catalyst reactor performance throughout the catalyst lifetime, considering catalyst deactivation. Furthermore, little attention has been given to the impact of various configurations of reactor networks and scheduling of the reactor operation (i.e., online and offline-regeneration) on the overall reactor performance throughout the catalyst lifetime. Therefore, we developed a range of feasible sequences of reactors and scheduling of reactors for operation and regeneration, and compared the overall reactor performance of multiple cases. Furthermore, a superstructure of reactor networks was developed and optimized to determine the optimum reactor network that shows the maximum overall reactor performance. The operating schedule of each reactor in the network was considered further. Lastly, the methodology was illustrated using a case study of the MTO (methanol to olefin) process.     

Alcoholic Fermentation with Self‐Flocculating Yeast in a Tower Upflow Reactor

Ethanol production using self‐flocculating yeast in a batch tower upflow reactor system operating with a recirculation loop was examined. Ethanol productivity, yield, and residual sucrose concentration were evaluated experimentally according to a central composite design with initial cell and sucrose concentrations and recirculation flow rate as independent variables. Yeast cell concentration strongly influenced the reactor performance. Alcoholic fermentation was conducted using this strain and reactor configuration which allowed for high productivity and high sucrose conversion. The ethanol yield was comparable with industrial yields. A kinetic study of the fermentation process under optimized conditions was performed using the experimental data and considering inhibition by sucrose and ethanol.     

Use of Neural Networks in the Simulation and Optimization of Pressure Swing Adsorption Processes

In this work a method for the simulation and optimization of a pressure swing adsorption process for the separation of nitrogen from air by using neural networks was developed. The model is used to obtain a prediction for the process performance, namely, the specific product and yield, over a wide range of operating conditions. These results are compared with the predictions from a mass tranfer model, and a very good agreement is found. The network developed is also used to minimize a cost objective function, and it is shown that it can easily be used in process optimization and/or control.     

翼型组合桨持气特性的实验研究及数据关联方法

研究了内径φ0.382～1.16m机械搅拌釜中翼型组合桨气液两相的持气特性,考察了结构参数 (包括翼型桨径、桨间距、桨下距离、通气位置、挡板形式及翼型桨排出流方向)和操作参数 (包括搅拌转速与通气量)对气含率的影响.采用单位体积功率和表观气速及Froude数和通气流动数两种方法对气含率进行了关联处理,并采用神经网络技术建立了翼型组合桨的气含率关联的网络模型,该模型掌握了各参数对气含率的影响规律,具有很好的泛化能力,与传统关联方法相比,网络模型的预测误差可缩小至±10％以内.     

醋酸胶溶剂对Hβ分子筛催化剂性能的影响

实验考察了醋酸胶溶剂对Hp分子筛催化剂成型强度的影响,并以丙烯和苯合成异丙苯为模型反应,分别以气-液半连续和液-液连续操作两种方式,实验评价催化剂的性能。结果表明：催化剂强度随成型过程中醋酸用量的增加呈现先增大后减小的趋势,最高催化剂强度对应的醋酸用量为6％。随着醋酸用量增加,催化剂活性略有增加,对异丙苯选择性影响不大,但催化剂稳定性却降低。     

ALKYLATION OF BENZENE USING BATCH AND CONTINUOUS FIXED-BED REACTORS

Alkylation reactions of benzene with propylene using heterogeneous catalysts H⁺-β zeolite, MCM-22, and ZSM-5 were studied for their affinity for cumene production. This work focused on the gas-phase reaction using different crystalline catalysts at several temperatures and amounts of reactants using both batch and continuous fixed-bed reactors. The properties of baseline commercial H⁺-β catalysts versus versions modified with Ga, La, and Pt were studied. Quantitative analysis of product mixture was performed by gas chromatography. For the batch reactor, β-zeolite produced the highest cumene yield and selectivity of 72% and 92%, respectively, at 225°C. At this temperature, a benzene:propylene dilution of 7:1 molar ratio was the optimum. For the continuous system, cumene production is favored at lower space velocities, higher benzene-to-propylene ratio, and temperatures close to 225°C. Ga modification of the H⁺-β zeolite significantly enhanced cumene yield in the continuous fixed-bed reactor at 225°C, from 27% of the unmodified β-zeolite to 36% for the Ga-modified one. The life span of modified β-catalysts was studied in the fixed-bed reactor for the first eight hours of reaction.     

离子液体催化烷基化体系在微反应器内的流动和反应基础研究

离子液体催化剂可以显著提高烷基化反应速率,开发适用于该工艺的高效反应器具有重要意义。基于微反应器在过程强化方面的优势,设计了适用于烷基化反应的微反应器,考察了烷基化模拟体系的流动规律,在较大操作范围内实现了催化剂与产物的连续分相。以上述研究为基础,实现了C₄烷基化反应的微型化,分别考察了停留时间、反应温度、分散尺寸对反应性能的影响。研究结果表明,微反应器内的烷基化反应在2 s内完成。C₈选择性和TMPs（三甲基戊烷）/DMHs（二甲基己烷）分别最高可达70.90%与13.4,均高于相同反应条件下搅拌釜反应器内的反应结果,证明微反应器在优化烷基化反应性能方面拥有巨大潜力。