基于差分进化算法的控制变量参数化方法及其在化工过程动态优化中的应用(英文)

Two general approaches are adopted in solving dynamic optimization problems in chemical processes, namely, the analytical and numerical methods. The numerical method, which is based on heuristic algorithms, has been widely used. An approach that combines differential evolution (DE) algorithm and control vector parameterization (CVP) is proposed in this paper. In the proposed CVP, control variables are approximated with polynomials based on state variables and time in the entire time interval. Region reduction strategy is used in DE to reduce the width of the search region, which improves the computing efficiency. The results of the case studies demonstrate the feasibility and efficiency of the proposed methods.     

降低精对苯二甲酸溶剂系统醋酸消耗的研究

Decreasing the acetic acid consumption in purified terephthalic acid （PTA） solvent system has become a hot issue with common concern. In accordance with the technical features, the electrical conductivity is in direct proportion to the acetic acid content. General regression neural network （GRNN） is used to establish the model of electrical conductivity on the basis of mechanism analysis, and then particle swarm optimization （PSO） algorithm with the improvement of inertia weight and population diversity is proposed to regulate the operating conditions. Thus, the method of decreasing the acid loss is derived and applied to PTA solvent system in a chemical plant. Cases studies show that the precision of modeling and optimization are higher. The results also provide the optimal operating conditions, which decrease the cost and improve the profit.     

在线最小二乘支持向量机及其在C8芳烃异构化建模中的应用

The least squares support vector regression (LS-SVR) is usually used for the modeling of single output system, but it is not well suitable for the actual multi-input-multi-output system. The paper aims at the modeling of multi-output systems by LS-SVR. The multi-output LS-SVR is derived in detail. To avoid the inversion of large matrix, the recursive algorithm of the parameters is given, which makes the online algorithm of LS-SVR practical. Since the computing time increases with the number of training samples, the sparseness is studied based on the projection of online LS-SVR. The residual of projection less than a threshold is omitted, so that a lot of samples are kept out of the training set and the sparseness is obtained. The standard LS-SVR, nonsparse online LS-SVR and sparse online LS-SVR with different threshold are used for modeling the isomerization of C8 aromatics. The root-mean-square-error (RMSE), number of support vectors and running time of three algorithms are compared and the result indicates that the performance of sparse online LS-SVR is more favorable.     

应用群能量恒定粒子群优化算法的批次流加发酵过程批次间优化(英文)

An iterative optimization strategy for fed-batch fermentation process is presented by combining a run-to-run optimization with swarm energy conservation particle swarm optimization (SEC-PSO). SEC-PSO, which is designed with the concept of energy conservation, can solve the problem of premature convergence frequently appeared in standard PSO algorithm by partitioning its population into several sub-swarms according to the energy of the swarm and is used in the optimization strategy for parameter iden-tification and operation condition optimization. The run-to-run optimization exploits the repetitive nature of fed-batch processes in order to deal with the optimal problems of fed-batch fermentation process with inaccurate process model and unsteady process state. The kinetic model parameters, used in the operation condition optimization of the next run, are adjusted by calculating time-series data obtained from real fed-batch process in the run-to-run optimization. The simulation results show that the strategy can adjust its kinetic model dynamically and overcome the instability of fed-batch process effectively. Run-to-run strategy with SEC-PSO provides an effective method for optimization of fed-batch fermentation process.     

化工过程系统的柔性分析扩展算法

An extended algorithm of flexibility analysis with a local adjusting method for flexibility region of chemical processes, which is based on the active constraint strategy, is proposed, which fully exploits the flexibility region of the process system operation. The hyperrectangular flexibility region determined by the extended algorithm is larger than that calculated by the previous algorithms. The limitation of the proposed algorithm due to imperfect convexity and its corresponding verification measure are also discussed. Both numerical and actual chemical process examples are presented to demonstrate the effectiveness of the new algorithm.     

A systematic approach for synthesizing a low-temperature distillation system

In this paper, by combining a stochastic optimization method with a refrigeration shaft work targeting method, an approach for the synthesis of a heat integrated complex distillation system in a low-temperature process is presented. The synthesis problem is formulated as a mixed-integer nonlinear programming (MINLP) problem, which is solved by simulated annealing algorithm under a random procedure to explore the optimal operating parameters and the distillation sequence structure. The shaft work targeting method is used to evaluate the minimum energy cost of the corresponding separation system during the optimization without any need for a detailed design for the heat exchanger network (HEN) and the refrigeration system (RS). The method presented in the paper can dramatical y reduce the scale and complexity of the problem. A case study of ethylene cold-end separation is used to il ustrate the application of the approach. Compared with the original industrial scheme, the result is encouraging.     

基于免疫系统的自适应差分进化算法及在水银氧化动力学参数估计中的应用

A new version of differential evolution (DE) algorithm, in which immune concepts and methods are applied to determine the parameter setting, named immune self-adaptive differential evolution (ISDE), is proposed to improve the performance of the DE algorithm. During the actual operation, ISDE seeks the optimal parameters arising from the evolutionary process, which enable ISDE to alter the algorithm for different optimization problems and improve the performance of ISDE by the control parameters’ self-adaptation. The performance of the proposed method is studied with the use of nine benchmark problems and compared with original DE algorithm and other well-known self-adaptive DE algorithms. The experiments conducted show that the ISDE clearly outperforms the other DE algorithms in all benchmark functions. Furthermore, ISDE is applied to develop the kinetic model for homogeneous mercury (Hg) oxidation in flue gas, and satisfactory results are obtained.     

一种改进的混合遗传算法及在化工过程动态优化中的应用(英文)

The solutions of dynamic optimization problems are usually very difficult due to their highly nonlinear and multidimensional nature. Genetic algorithm (GA) has been proved to be a feasible method when the gradient is difficult to calculate. Its advantage is that the control profiles at all time stages are optimized simultaneously, but its convergence is very slow in the later period of evolution and it is easily trapped in the local optimum. In this study, a hybrid improved genetic algorithm (HIGA) for solving dynamic optimization problems is proposed to overcome these defects. Simplex method (SM) is used to perform the local search in the neighborhood of the optimal solution. By using SM, the ideal searching direction of global optimal solution could be found as soon as possible and the convergence speed of the algorithm is improved. The hybrid algorithm presents some improvements, such as protecting the best individual, accepting immigrations, as well as employing adaptive crossover and Gaussian mutation operators. The efficiency of the proposed algorithm is demonstrated by solving several dynamic optimization problems. At last, HIGA is applied to the optimal production of secreted protein in a fed batch reactor and the optimal feed-rate found by HIGA is effective and relatively stable.     

基于符号有向图与定性趋势分析的化工过程故障诊断方法(英文)

In the past 30 years, signed directed graph (SDG), one of the qualitative simulation technologies, has been widely applied for chemical fault diagnosis. However, SDG based fault diagnosis, as any other qualitative method, has poor diagnostic resolution. In this paper, a new method that combines SDG with qualitative trend analysis (QTA) is presented to improve the resolution. In the method, a bidirectional inference algorithm based on assumption and verification is used to find all the possible fault causes and their corresponding consistent paths in the SDG model. Then an improved QTA algorithm is used to extract and analyze the trends of nodes on the consistent paths found in the previous step. New consistency rules based on qualitative trends are used to find the real causes from the candidate causes. The resolution can be improved. This method combines the completeness feature of SDG with the good diagnostic resolution feature of QTA. The implementation of SDG-QTA based fault diagnosis is done using the integrated SDG modeling, inference and post-processing software platform. Its application is illustrated on an atmospheric distillation tower unit of a simulation platform. The result shows its good applicability and efficiency.     

Integration of a Gas Fired Steam Power Plant with a Total Site Utility Using a New Cogeneration Targeting Procedure

A steam power plant can work as a dual purpose plant for simultaneous production of steam and elec-trical power. In this paper we seek the optimum integration of a steam power plant as a source and a site utility sys-tem as a sink of steam and power. Estimation for the cogeneration potential prior to the design of a central utility system for site utility systems is vital to the targets for site fuel demand as well as heat and power production. In this regard, a new cogeneration targeting procedure is proposed for integration of a steam power plant and a site utility consisting of a process plant. The new methodology seeks the optimal integration based on a new cogenera-tion targeting scheme. In addition, a modified site utility grand composite curve （SUGCC） diagram is proposed and compared to the original SUGCC. A gas fired steam power plant and a process site utility is considered in a case study. The applicability of the developed procedure is tested against other design methods （STAR？ and Thermoflex software） through a case study. The proposed method gives comparable results, and the targeting method is used for optimal integration of steam levels. Identifying optimal conditions of steam levels for integration is important in the design of utility systems, as the selection of steam levels in a steam power plant and site utility for integration greatly influences the potential for cogeneration and energy recovery. The integration of steam levels of the steam power plant and the site utility system in the case study demonstrates the usefulness of the method for reducing the overall energy consumption for the site.     

水煤浆气化装置操作优化技术及其应用

针对水煤浆气化装置操作优化问题,提出了一种多种群竞争型协同文化差分进化算法(MCCDE),算法中建立了基于差分进化算法的竞争型协同策略及竞争适应度评判方法,并引入了文化算法的部分思想。同时,建立了德士古气化炉操作优化模型,将MCCDE算法用于优化模型的求解。采用某化工厂气化系统实际运行数据进行仿真,经过操作优化计算,能够获得优化的控制参数,并提高气化炉有效气产率。最后,开发了水煤浆气化操作优化系统应用软件,能够实现将建模、控制、优化技术应用于实际生产中,以提高装置的经济效益。     

广义回归神经网络的改进及在延迟焦化建模中的应用

广义回归神经网络（GRNN）具有明确的概率意义,其参数大多能自动确定,仅光滑因子参数需优化估值.采用优进遗传算法（EGA）,将确定性与随机性寻优操作相融合,实现了高效全局搜优,它所基于的优进策略包括设计Powell寻优算子、改进交叉算子、自适应地调整交叉率和变异率等.以推广能力作为优化目标,所建的GRNN有很强的非线性拟合能力和优良的预报性能,将其成功地为延迟焦化过程建模,与径向基网络（RBFN）等相比,显示了明显的优势.     

Optimization of fractionation operation in the benzene-toluene-xylene plant

A steady-state optimization for the fractionation unit of an industrial benzene-toluene-xylene (BTX) plant was developed. Because the fractionation unit of the BTX plant has narrow ranges of boiling point and doesn’t have any sidecut and side reboiler, we employed the boiling point estimation method in the modeling and simulation of the unit. The well-known Soave-Redlich-Kwong state equation was employed to compute required thermodynamic properties. The results of simulations showed very good agreement with the actual operation data. The objective of the optimization was to maximize profit of fractionation unit. The successive quadratic programming algorithm was utilized and objective functions for each fractionation column were set up based on utility and operation costs and selling prices of products. The graphical user interface of the optimization system was developed to provide convenience and flexibility in actual applications.     

基于鲸鱼优化算法的汽轮机热耗率模型预测

为了准确地建立汽轮机热耗率预测模型,提出了一种基于反向学习自适应的鲸鱼优化算法（AWOA）和快速学习网（FLN）综合建模的方法。首先将改进后的鲸鱼算法与经典改进的粒子群、差分进化算法和基本鲸鱼算法进行比较,结果证明其具有更高的收敛精度和更快的收敛速度;然后采用某热电厂600 MW超临界汽轮机组现场收集的运行数据建立汽轮机热耗率预测模型,并将改进后的鲸鱼算法优化的快速学习网模型的预测结果与基本快速学习网及经典改进的粒子群、差分进化算法和基本鲸鱼算法优化的快速学习网模型预测结果相比较。结果表明,AWOA-FLN预测模型具有更高的预测精度和更强的泛化能力,更能准确地预测汽轮机的热耗率。     

基于改进差分进化和最小二乘支持向量机的铝酸钠溶液浓度软测量

针对氧化铝蒸发过程铝酸钠溶液浓度难以在线检测问题,提出了改进差分进化和最小二乘支持向量机的铝酸钠溶液浓度软测量建模方法。首先基于灰色关联分析和核主成分分析确定模型的输入变量,再用改进差分进化算法的最小二乘支持向量机构建软测量模型。并与DE-LSSVM软测量模型进行比较;最后应用蒸发过程生产数据进行验证,结果表明,新模型具有更好的学习能力和泛化性能且预测精度更高,可为蒸发过程操作优化提供必要的指导。     

基于NSGA-Ⅱ的模拟移动床色谱分离过程多目标操作优化

多目标优化策略被应用于模拟移动床过程的操作优化中，采用一种基于Pareto最优解的多目标优化算法——NSGA-Ⅱ算法，以分离联萘酚对映体的模拟移动床色谱分离过程作为研究对象，利用模拟移动床TMB数学模型，以分离性能指标作为目标函数进行了多目标操作优化设计。优化结果表明，NSGA-Ⅱ算法得到的非劣解在目标空间分布均匀，算法收敛性和鲁棒性好。基于NSGA-Ⅱ算法的面向分离性能多目标优化设计方法为模拟移动床分离过程的工艺设计和操作指导提供了有效的工具。     

Optimal operation of utility systems in petrochemical plants

Several operational modes can be used in the operation of utility plants for petrochemical plants. For the optimization system of a utility plant to be effective, all the possible operational modes should be taken into account. Moreover, due to the variable fuel cost and electricity cost, the objective function in the optimization system may take different forms depending on the utility management strategy. In this paper, we present a utility optimization system that is based on mixed integer linear programming and considers all possible operational modes and various types of objective functions to maximize the flexibility and the usefulness of the optimization system. The user can conveniently choose a suitable operation mode and a type of the objective function. Results of the optimization can be displayed both numerically and graphically.     

深入探索智能算法与反应网络研究的融合

反应网络是化工过程机理在微观分子尺度上的表达方式，但网络的复杂性为深入认识生产过程提出了挑战。本文提出了探索智能算法与反应网络研究融合的思路，基于物质转化的“透明工程”的概念，深入剖析反应网络的结构统计指标、结构拓扑特征、节点性质特征、机理动态演化、建模应用性能等特点。随后阐述了使用“数据结构化、智能优化与分析、智能代理建模”三步结合的机理数值化反应网络研究方法，既实现了在微观层面的局部放大，又实现了在工业应用中的准确预测。文中指出，智能算法融合反应网络后可以对实际工业过程执行可视化、可解释性的建模、分析与优化，为相关工业生产提质增效提供决策依据，并进一步帮助人类突破认知的极限，更深入地理解反应过程，提取关键的反应规律，助力化学工业的智能化。     

基于Kriging的差分进化算法及其在苯乙烯流程优化中的应用

自适应差分进化算法基于个体生成策略和控制参数自适应,无须人为设置参数,对问题有较好的适应性,但其收敛速度和精度有待提高。将具有较高预测精度的Kriging模型应用于自适应差分进化算法中,建立跟随种群变化的Kriging模型,通过模型极值点与种群最优个体竞争,对种群产生扰动,影响种群进化过程,改善算法的收敛速度和寻优性能。对10个典型测试函数的测试结果表明,该算法较标准和自适应差分进化算法收敛速度加快,收敛精度提高,且具有更好的稳定性。将基于Kriging的差分进化算法应用于苯乙烯装置的流程优化,操作运行费用显著降低。     

基于自适应模糊神经网络的非线性系统模型预测控制

针对非线性动态系统的控制问题,提出了一种基于自适应模糊神经网络(adaptive fuzzy neural network, AFNN)的模型预测控制(model predictive control, MPC)方法。首先,在离线建模阶段,AFNN采用规则自分裂技术产生初始模糊规则,采用改进的自适应LM学习算法优化网络参数;然后,在实时控制过程,AFNN根据系统输出和预测输出之间的误差调整网络参数,从而为MPC提供一个精确的预测模型;进一步,AFNN-MPC利用带有自适应学习率的梯度下降寻优算法求解优化问题,在线获取非线性控制量,并将其作用到动态系统实施控制。此外,给出了AFNN-MPC的收敛性和稳定性证明,以保证其在实际工程中的成功应用。最后,利用数值仿真和双CSTR过程进行实验验证。结果表明,AFNN-MPC能够取得优越的控制性能。