催化剂配方的神经网络建模与遗传算法优化

采用神经网络对催化剂配方实验数据建模,在网络模型仿真实验的基础上应用遗传算法对配方组成进行优化,得到的优化配方经实验验证,证实了该算法的实际应用意义.该算法对化工产品的配方及工艺优化也有实用意义.     

基于EBF神经网络和粒子群算法的注射成型优化设计

基于拉丁超立方设计建立了椭球基（EBF）神经网络模型描述注塑工艺参数与翘曲值间的函数关系,将EBF神经网络模型与Kriging模型对比,说明EBF神经网络模型可以准确地描述注塑工艺参数与翘曲值之间的函数关系,并结合多目标粒子群算法对工艺参数进行优化,并与邻域培植遗传算法优化结果对比,说明多目标粒子群算法的优点。结果表明,基于EBF神经网络模型和粒子群优化算法可以使塑料出水管翘曲值减小11.64 %,同时使保压时间和冷却时间总和减小了2.13 s,从而在出水管批量生产过程中减少了生产时间。     

基于智能集成优化的合成塔入口氨含量软测量

文化算法主要由种群空间和信念空间构成,两个空间共同进化来模拟人类文化的进化过程。提出一种智能集成优化算法,将遗传算法、粒子群优化算法应用到文化算法框架中,并与神经网络相结合,构造一种智能集成网络模型,并将其应用到合成塔入口氨含量软测量建模。结合实际工艺,对所建软测量模型进行仿真研究。研究结果表明,该模型的性能优于传统BP神经网络模型、遗传神经网络模型和粒子群神经网络模型,具有较高的精度和良好的应用前景。     

粒子群优化小波神经网络的齿轮箱故障诊断

针对传统的小波神经网络容易陷入局部极小、收敛速度慢和运行效率低等不足,将粒子群优化引入小波神经网络中,提出了一种粒子群优化小波神经网络学习算法。该方法利用粒子群算法优化小波神经网络中的权值(即尺度因子)和阈值(即平移因子),从而可以自适应地选择小波神经网络的参数,提高算法的收敛性和快速性,克服传统的小波神经网络容易陷入局部极小、收敛速度慢和运行效率低等不足。实验结果表明,提出的方法是有效的,优于传统的小波神经网络学习算法,能够准确地识别齿轮裂纹的损坏程度。     

两群微粒群算法及其在油品调和优化中的应用

针对复杂的非线性约束优化问题,提出了一种含变异算子的两群微粒群算法。算法构造了两个粒子群,分别设置了不同的搜索速度上限,并设计了粒子群间的协调机制和变异算子,使算法的寻优能力得到增强。针对油品调和配方优化进行了实例仿真,研究结果表明所提出的算法能获得较理想的调和配方,在满足调和利润最大的同时能保证对调和指标的卡边,使调和成品油的指标富余量大大降低。     

基于粒子群神经网络的异步电机故障诊断系统的研究

采用粒子群算法与BP算法结合的综合方式训练神经网络,用粒子群优化算法调节和优化全局性的网络参数,用BP神经网络学习方法优化局部性的参数,解决单纯的BP算法训练收敛较慢且易陷入局部最小值等问题,用于提高异步电机故障诊断的性能。在此基础上借助MATLAB进行计算机程序的编写,并使用Visual Basic进行异步电机故障诊断系统用户操作界面的开发。实践证明该系统能有效识别电机常见故障,操作简单方便且诊断精度高。     

陶瓷坯釉料配方优化算法及其实现

陶瓷坯釉料配方最优化计算是陶瓷工艺中的关键技术之一。本文利用复合形优化算法选出最佳的配方比例,通过加权值的调整,使得陶瓷坯釉料配方结果完全满足优化目标。     

粒子群优化算法的发展及应用

讨论了粒子群优化算法的发展和应用。介绍了粒子群优化算法的基本原理和算法流程,并且与其他演化算法进行了比较,给出了一些经常用到的测试函数。针对粒子群优化算法在搜索后期存在的不足,介绍了改进的粒子群优化算法,重点介绍了在实际应用领域中用到的改进粒子群优化算法。     

余热锅炉单相受热面动态建模与模型参数优化

作为余热利用环节中最重要的部分,余热锅炉的启动、变工况运行和停机特性直接决定锅炉的寿命及效率。基于工质热力学性质和质量、动量及能量守恒方程,以Matlab/Simulink为平台,构建了余热锅炉单相受热面的动态仿真模型。结合某水泥厂自主设计的直流余热锅炉实验数据,基于遗传算法和粒子群算法,对动态模型进行了参数优化。结果表明,经过优化后,余热锅炉动态模型与实验数据匹配程度高,模拟与实验结果的误差为0.93%~4.39%。因此,本文所建立的单相受热面变工况动态模型可以准确反映余热锅炉受热面动态特性。两种算法的对比表明,粒子群算法适应度函数收敛更优;在收敛迭代次数上,粒子群算法在54~64代达到收敛,遗传算法在93代后达到收敛。粒子群算法在参数优化方面优于遗传算法。     

粒子群优化神经网络的电镀金刚石套钻使用寿命预测

电镀金刚石套钻广泛应用各种加工领域,而套钻使用寿命受到多种因素影响,具有一定的非线性变化特点,为了准确预测电镀金刚石套钻的使用寿命,提出了基于粒子群优化神经网络的电镀金刚石套钻使用寿命预测模型。对电镀金刚石套钻使用寿命预测现状进行分析,针对BP神经网络参数优化问题,采用粒子群优化算法确定最优参数,建立电镀金刚石套钻使用寿命的预测模型,通过仿真实验对其有效性和优越性进行分析。实验结果表明,模型可以准确刻画影响参数与电镀金刚石使用寿命之间的变化关系,获得比对比模型更高的预测精度,实际应用价值更高。     

微粒群优化算法研究进展

围绕微粒群优化(PSO)算法的原理、特点、改进、应用等方面进行全面综述,介绍针对复杂环境的PSO研究内容,包括多目标、约束、离散和动态优化等,提出PSO有待进一步研究的若干方向和内容。     

Model-free real-time optimization of process systems using safe Bayesian optimization

Conventional real-time optimization (RTO) requires detailed process models, which may be challenging or expensive to obtain. Model-free RTO methods are an attractive alternative to circumvent the challenge of developing accurate models. Most model-free RTO methods are based on estimating the steady-state cost gradient with respect to the decision variables and driving the estimated gradient to zero using integral action. However, accurate gradient estimation requires clear time scale separation from the plant dynamics, such that the dynamic plant can be assumed to be a static map. For processes with long settling times, this can lead to prohibitively slow convergence to the optimum. To avoid the need to estimate the cost gradients from the measurement, this article uses Bayesian optimization, which is a zeroth order black-box optimization framework. In particular, this article uses a safe Bayesian optimization based on interior point methods to ensure that the setpoints computed by the model-free steady-state RTO layer are guaranteed to be feasible with high probability (i.e., the safety-critical constraints will not be violated at steady-state). The proposed method can thus be seen as a model-free variant of the conventional two-step steady-state RTO framework (with steady-state detection), which is demonstrated on a benchmark Williams-Otto reactor example.     

福建乙烯装置裂解炉优化运行概述

根据裂解炉开工后实际运行情况,从控制、原料、运行模式和管理等方面对裂解炉运行进行优化,实现了新建乙烯装置裂解炉安全、平稳和长周期运行。结果发现：通过不断调整测试和优化,裂解炉实现了100％自控率;依据原料性质,停止裂解重石脑油,将重石脑油全部改供给小重整装置做原料,同时将碳四与轻石脑油混和,提高了产率与原料利用率;采用“7＋l”裂解模式并进行在线烧焦大大降低了运行成本。     

流量同时作为工艺优化变量和串级控制副变量时的工艺优化实现与分析

针对如何通过现存的"主被控变量-流量"这种典型的化工过程串级控制实现流量优化的在线优化问题,对在线优化的实现结构进行分析,给出一种流量优化变量通过一阶惯性环节改变主回路被控变量设定值的在线优化的理想方案,通过CSTR的仿真实例证明了优化实现方案的可行性。     

HEURISTIC FEED RATE PROFILES FOR OPTIMAL YIELD AND PRODUCTIVITY OF FED-BATCH BIOPROCESSES

The optimization of two important classes of fed-batch fermentation processes with respect to the limiting substrate feed rate is considered: (i) growth of one biomass on one limiting substrate, and (ii) growth and product formation on one substrate. For these two types of bioprocesses, two different performance criteria are compared: yield and productivity. Heuristic Teed rate profiles are derived based on (i) the optimal control solution according to the minimum principle, and (ii) the qualitative behavior of the specific rates for growth and possibly product formation. For biomass production processes, it is illustrated that the heuristic strategy coincides exactly with the optimal control solution, while its structure is independent of the performance index considered. For product formation processes, it is illustrated that the same heuristic approach can be used to optimize both yield and productivity in spite of the well known conflict between them.     

A new approach for global optimization of a class of MINLP problems with applications to water management and pooling problems

One of the biggest challenges in solving optimization engineering problems is rooted in the nonlinearities and nonconvexities, which arise from bilinear terms corresponding to component material balances and/or concave functions used to estimate capital cost of equipments. The procedure proposed uses an MILP lower bound constructed using partitioning of certain variables, similar to the one used by other approaches. The core of the method is to bound contract a set of variables that are not necessarily the ones being partitioned. The procedure for bound contraction consists of a novel interval elimination procedure that has several variants. Once bound contraction is exhausted the method increases the number of intervals or resorts to a branch and bound strategy where bound contraction takes place at each node. The procedure is illustrated with examples of water management and pooling problems. © 2011 American Institute of Chemical Engineers AIChE J, 58: 2320–2335, 2012     

A global optimization method based on multi-unit extremum-seeking for scalar nonlinear systems

Finding the global optimum of a nonlinear function is a challenging task that could involve a large number of functional evaluations. In this paper, an algorithm that uses tools from the domain of extremum-seeking is shown to provide an efficient deterministic method for global optimization. Extremum-seeking schemes typically find the local optimum by controlling the gradient to zero. In this paper, the multi-unit framework is used, where the gradient is estimated by finite difference for a given offset between the inputs. The gradient is pushed to zero by an integral controller. It is shown that if the offset is reduced to zero, the system can be made to converge to the global optimum of nonlinear continuous static, scalar maps. The result is extended to constrained problems where a switching control strategy is employed. Several illustrative examples are presented and the proposed method is compared with other methods of global optimization.     

Targeting climate-neutral hydrogen production: Integrating brown and blue pathways with green hydrogen infrastructure via a novel superstructure and simulation-based life cycle optimization

This article addresses the sustainable design of hydrogen (H₂) production systems that integrate brown and blue pathways with green hydrogen infrastructure. We develop a systematic framework to simultaneously optimize the process superstructure and operating conditions of steam methane reforming (SMR)-based hydrogen production systems. A comprehensive superstructure that integrates SMR with multiple carbon dioxide capture technologies, electrolyzers, fuel cells, and working fluids in the organic rankine cycle is proposed under varying operating conditions. A life cycle optimization model is then developed by integrating superstructure optimization, life cycle assessment approach, techno-economic assessment, and process optimization using extensive process simulation models and formulated as a mixed-integer nonlinear program. We find that the optimal unit-levelized cost of hydrogen ranges from $1.49 to $3.18 per kg H₂. Moreover, the most environmentally friendly process attains net-zero life cycle greenhouse gas emissions compared to 10.55 kg CO₂-eq per kg H₂ for the most economically competitive process design.     

Empirical strategies for open-loop on-line optimization

Two empirical strategies for open-loop on-line optimization are developed as alternatives to the use of mechanistic process models. These strategies are based on on-line identification of dynamic multi-input single-output (MISO) and multi-input multi-output (MIMO) models. The steady state gain of these models provides information for steady state optimization. Desirability functions, originally developed for multi-objective optimization, are utilized as objective function modifiers for constrained on-line optimization. The integration of dynamic model identification and desirability functions results in an on-line optimizer which combines fast optimizing speed with the ability to predict future encroachments on constraint boundaries. Corrections to the search direction are based on these predictions, reducing the probability of actual constraint violation. The optimization strategies are tested by simulation on nonlinear multivariable interacting systems at two levels of complexity: a CSTR supporting a multiple reaction and a fluid catalytic cracker. Both methods were effective in avoiding violation of constraints but the MIMO strategy required fewer steps to reach an optimum and was less prone to generate a nonfeasible optimization step.     

ROBUST MEASUREMENT-BASED OPTIMIZATION WITH STEADY-STATE DIFFERENTIAL EQUATION MODELS

A robust model-based iterative feedback optimization methodolgy has been introduced in previous work for the steady-state optimization of chemical process operations without requiring cumbersome model updating. Here we extend the methodology to utilize differential equation models directly. The adjoint system method that is often used in dynamic optimization is modified to explicitly incorporate plant measurements in the gradient computation during each iteration, resulting in robustness with respect to model-plant mismatch. In the case that the states cannot be measured along the entire spatial direction, estimated profiles based on boundary measurements are utilized. The methodology is tested with simulations of an ammonia synthesis reactor in autothermal operation and is shown to be robust in the presence of modeling error, input error, and measurement noise.