Computers & Chemical Engineering最新文献_第8页

C2 and C+ compound production from carbon dioxide: supply chain design and optimization 从二氧化碳中生产C2和C+化合物：供应链设计和优化

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-31 DOI: 10.1016/j.compchemeng.2025.109476

Grazia Leonzio , Nilay Shah

The use of carbon dioxide to produce chemical products can in principle decrease both fossil resource exploitation and greenhouse gas emissions. Potential compounds that can be obtained are ethylene and polyurethane, characterized by large markets underpinning many industries. The importance of such platform chemicals means that different routes have been investigated in the literature for their alternative synthesis. However, there is lack of a direct comparison of these products and processes in a carbon capture utilization and storage supply chain in the existing literature: a gap which we wish to address here.

A mixed integer linear optimization model for a carbon capture utilization and storage supply chain is developed here: carbon dioxide can be extracted from flue gas to be stored and/or utilized for ethylene (via the tandem or methanol to olefin processes), polyethylene or polyurethane production and these can be either sold immediately or stored. The framework is exemplified by a case study localised within the UK Teesside petrolchemical cluster and the best topology is suggested to minimize total costs. Moreover, a dynamic analysis of the system over the years is considered here to suggest the best way to implement carbon dioxide capture and utilisation through to 2050. Results show that the optimal cost is achieved by capturing carbon dioxide and converting it into ethylene via the methanol-to-olefin process; the whole system levelized cost is 7.3 $/kg_Ethylene. Moreover, ensuring an homogeneous way of capturing carbon dioxide over time maximises the profitability of the overall system.

利用二氧化碳生产化学产品原则上可以减少化石资源的开采和温室气体的排放。可以获得的潜在化合物是乙烯和聚氨酯，其特点是支撑许多行业的巨大市场。这些平台化学物质的重要性意味着文献中已经研究了不同的替代合成途径。然而，在现有文献中缺乏对碳捕获利用和储存供应链中这些产品和过程的直接比较：我们希望在这里解决这一差距。这里开发了一个用于碳捕获利用和储存供应链的混合整数线性优化模型：可以从烟气中提取二氧化碳，用于储存和/或用于乙烯（通过串联或甲醇制烯烃工艺）、聚乙烯或聚氨酯生产，这些可以立即出售或储存。该框架以英国Teesside石化集群的案例研究为例，并建议最佳拓扑结构以最大限度地降低总成本。此外，通过对该系统多年来的动态分析，本文提出了到2050年实现二氧化碳捕获和利用的最佳方法。结果表明，采用甲醇制烯烃工艺捕集二氧化碳并将其转化为乙烯的成本最优；整个系统平准化成本为7.3美元/ kge乙烯。此外，确保一种随时间推移捕获二氧化碳的均匀方式，将使整个系统的盈利能力最大化。

{"title":"C2 and C+ compound production from carbon dioxide: supply chain design and optimization","authors":"Grazia Leonzio , Nilay Shah","doi":"10.1016/j.compchemeng.2025.109476","DOIUrl":"10.1016/j.compchemeng.2025.109476","url":null,"abstract":"<div><div>The use of carbon dioxide to produce chemical products can in principle decrease both fossil resource exploitation and greenhouse gas emissions. Potential compounds that can be obtained are ethylene and polyurethane, characterized by large markets underpinning many industries. The importance of such platform chemicals means that different routes have been investigated in the literature for their alternative synthesis. However, there is lack of a direct comparison of these products and processes in a carbon capture utilization and storage supply chain in the existing literature: a gap which we wish to address here.</div><div>A mixed integer linear optimization model for a carbon capture utilization and storage supply chain is developed here: carbon dioxide can be extracted from flue gas to be stored and/or utilized for ethylene (via the tandem or methanol to olefin processes), polyethylene or polyurethane production and these can be either sold immediately or stored. The framework is exemplified by a case study localised within the UK Teesside petrolchemical cluster and the best topology is suggested to minimize total costs. Moreover, a dynamic analysis of the system over the years is considered here to suggest the best way to implement carbon dioxide capture and utilisation through to 2050. Results show that the optimal cost is achieved by capturing carbon dioxide and converting it into ethylene via the methanol-to-olefin process; the whole system levelized cost is 7.3 $/kg<sub>Ethylene</sub>. Moreover, ensuring an homogeneous way of capturing carbon dioxide over time maximises the profitability of the overall system.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109476"},"PeriodicalIF":3.9,"publicationDate":"2025-10-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145576225","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Decarbonization of the electrical sector: A new model for the environmental assessment of global copper cathode and aluminum ingot supply chains 电力部门的脱碳：全球铜阴极和铝锭供应链环境评估的新模型

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-30 DOI: 10.1016/j.compchemeng.2025.109475

Andreia Santos, Adrian Rodriguez, Ana Carvalho

Achieving carbon neutrality and reducing emissions has been mainly achieved through the decarbonization of the electricity sector. Moving towards an electric sector primarily based on renewables can substantially lower greenhouse gas emissions and decrease dependence on fossil fuel. However, this energy transition comes at a cost. Many raw materials, particularly minerals like copper and aluminum, are required for the transition, but the extraction, beneficiation, and processing of these minerals into uniform products, such as copper cathodes and aluminum ingots, comes with environmental burdens that should be addressed. Literature has some specific environmental models to assess copper cathodes and aluminum ingots production, however these models do not represent the global copper cathode and aluminum ingot supply chains and do not allow to compare both processed minerals. Therefore, this paper aims to propose a new model of the global copper cathode and aluminum ingot supply chains, which allows the diagnosis in environmental terms of the two global supply chains and allows their comparison in environmental terms. The Life Cycle Assessment methodology was employed to create the model and to assess the environmental impacts and SimaPro software was used to implement the model. Results show that, for copper cathodes, human non-carcinogenic toxicity is the dominant contributor to the overall impact of the global supply chain, largely driven by arsenic and lead emissions during copper anode production. In contrast, for aluminum ingots, fine particulate matter formation, mainly linked to electricity consumption during smelting, emerges as the most significant environmental burden. The comparison between the two global supply chains revealed that the total environmental burden associated with producing 1 ton of copper cathodes is nearly four times higher than that of producing the same amount of aluminum ingots. These results are useful for policy-makers, when establishing decarbonization strategies and policies. It will be also useful for producers to decrease their environmental burden. Finally, producers will also be able to select the greenest metal for their production, when copper and aluminum can work as substitutes.

实现碳中和和减少排放主要是通过电力部门的脱碳实现的。向以可再生能源为基础的电力行业发展，可以大幅降低温室气体排放，减少对化石燃料的依赖。然而，这种能源转型是有代价的。转型需要许多原材料，特别是铜和铝等矿物，但这些矿物的提取、选矿和加工成统一的产品，如铜阴极和铝锭，会带来环境负担，应该得到解决。文献中有一些特定的环境模型来评估铜阴极和铝锭的生产，但是这些模型并不代表全球铜阴极和铝锭的供应链，也不允许比较这两种加工矿物。因此，本文旨在提出一种新的全球铜阴极和铝锭供应链模型，该模型允许在环境方面对两个全球供应链进行诊断，并允许在环境方面对它们进行比较。采用生命周期评估方法创建模型并评估环境影响，并使用SimaPro软件实现模型。结果表明，对于铜阴极，人类非致癌毒性是全球供应链总体影响的主要贡献者，主要由铜阳极生产过程中的砷和铅排放驱动。相比之下，对于铝锭，细颗粒物的形成主要与冶炼过程中的电力消耗有关，成为最严重的环境负担。两个全球供应链之间的比较显示，生产1吨铜阴极的总环境负担几乎是生产相同数量铝锭的四倍。这些结果对决策者在制定脱碳战略和政策时是有用的。这也将有助于生产者减轻其环境负担。最后，当铜和铝可以作为替代品时，生产商也将能够选择最环保的金属来生产。

{"title":"Decarbonization of the electrical sector: A new model for the environmental assessment of global copper cathode and aluminum ingot supply chains","authors":"Andreia Santos, Adrian Rodriguez, Ana Carvalho","doi":"10.1016/j.compchemeng.2025.109475","DOIUrl":"10.1016/j.compchemeng.2025.109475","url":null,"abstract":"<div><div>Achieving carbon neutrality and reducing emissions has been mainly achieved through the decarbonization of the electricity sector. Moving towards an electric sector primarily based on renewables can substantially lower greenhouse gas emissions and decrease dependence on fossil fuel. However, this energy transition comes at a cost. Many raw materials, particularly minerals like copper and aluminum, are required for the transition, but the extraction, beneficiation, and processing of these minerals into uniform products, such as copper cathodes and aluminum ingots, comes with environmental burdens that should be addressed. Literature has some specific environmental models to assess copper cathodes and aluminum ingots production, however these models do not represent the global copper cathode and aluminum ingot supply chains and do not allow to compare both processed minerals. Therefore, this paper aims to propose a new model of the global copper cathode and aluminum ingot supply chains, which allows the diagnosis in environmental terms of the two global supply chains and allows their comparison in environmental terms. The Life Cycle Assessment methodology was employed to create the model and to assess the environmental impacts and SimaPro software was used to implement the model. Results show that, for copper cathodes, human non-carcinogenic toxicity is the dominant contributor to the overall impact of the global supply chain, largely driven by arsenic and lead emissions during copper anode production. In contrast, for aluminum ingots, fine particulate matter formation, mainly linked to electricity consumption during smelting, emerges as the most significant environmental burden. The comparison between the two global supply chains revealed that the total environmental burden associated with producing 1 ton of copper cathodes is nearly four times higher than that of producing the same amount of aluminum ingots. These results are useful for policy-makers, when establishing decarbonization strategies and policies. It will be also useful for producers to decrease their environmental burden. Finally, producers will also be able to select the greenest metal for their production, when copper and aluminum can work as substitutes.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109475"},"PeriodicalIF":3.9,"publicationDate":"2025-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145463217","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Metrics for evaluating stochastic outputs in machine learning models: Addressing accuracy and uncertainty 评估机器学习模型中随机输出的度量：处理准确性和不确定性

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-29 DOI: 10.1016/j.compchemeng.2025.109474

Yushi Deng, Mario R. Eden, Selen Cremaschi

Chemical processes are modeled and designed considering uncertainties in system parameters, operating conditions, and environmental factors. Models with stochastic outputs are commonly employed to make decisions in chemical engineering designs. Evaluating models with stochastic outputs requires assessing both prediction accuracy and precision. The area metric measures the overall mismatch between the prediction and observation, but does not provide a metric to assess precision and accuracy separately. Previously, we introduced uncertainty width, which decomposes the area metric into precision and bias components for model outputs whose distributions are symmetric. In this work, we investigate the applicability and effectiveness of the uncertainty width to asymmetric output distributions through a series of computational experiments. We then further study the application of the uncertainty width to evaluate the performance of eight distinct machine learning techniques, each integrated within a hybrid modeling framework, for predicting liquid entrainment with its uncertainty across three different flow orientations. The results of the computational experiments suggest that the uncertainty width is effective for asymmetric cases, especially when bias is small. Further studies are needed to understand the effectiveness of uncertainty width in cases of large bias and extreme asymmetry. The results for the hybrid models support the effectiveness of uncertainty width. They reveal that the Gaussian Process model has the best overall prediction accuracy. The remaining models exhibit diverse trade-offs between precision and accuracy, indicating that model selection should be guided by the specific accuracy and precision requirements for the application.

化学过程的建模和设计考虑了系统参数、操作条件和环境因素的不确定性。具有随机输出的模型通常用于化学工程设计决策。评估具有随机输出的模型需要评估预测的准确性和精度。面积度量度量预测和观测之间的总体不匹配，但不提供单独评估精度和准确性的度量。在此之前，我们引入了不确定性宽度，它将分布对称的模型输出的面积度量分解为精度和偏差分量。在这项工作中，我们通过一系列的计算实验来研究不确定性宽度对非对称输出分布的适用性和有效性。然后，我们进一步研究了不确定性宽度的应用，以评估八种不同的机器学习技术的性能，每种技术都集成在一个混合建模框架中，用于预测三种不同流动方向的液体夹带及其不确定性。计算实验结果表明，不确定度宽度对不对称情况是有效的，特别是当偏差较小时。需要进一步的研究来了解不确定宽度在大偏差和极端不对称情况下的有效性。混合模型的结果支持了不确定性宽度的有效性。结果表明，高斯过程模型具有最佳的整体预测精度。其余的模型在精度和精度之间表现出不同的权衡，表明模型选择应该由应用程序的特定精度和精度要求来指导。

{"title":"Metrics for evaluating stochastic outputs in machine learning models: Addressing accuracy and uncertainty","authors":"Yushi Deng, Mario R. Eden, Selen Cremaschi","doi":"10.1016/j.compchemeng.2025.109474","DOIUrl":"10.1016/j.compchemeng.2025.109474","url":null,"abstract":"<div><div>Chemical processes are modeled and designed considering uncertainties in system parameters, operating conditions, and environmental factors. Models with stochastic outputs are commonly employed to make decisions in chemical engineering designs. Evaluating models with stochastic outputs requires assessing both prediction accuracy and precision. The area metric measures the overall mismatch between the prediction and observation, but does not provide a metric to assess precision and accuracy separately. Previously, we introduced uncertainty width, which decomposes the area metric into precision and bias components for model outputs whose distributions are symmetric. In this work, we investigate the applicability and effectiveness of the uncertainty width to asymmetric output distributions through a series of computational experiments. We then further study the application of the uncertainty width to evaluate the performance of eight distinct machine learning techniques, each integrated within a hybrid modeling framework, for predicting liquid entrainment with its uncertainty across three different flow orientations. The results of the computational experiments suggest that the uncertainty width is effective for asymmetric cases, especially when bias is small. Further studies are needed to understand the effectiveness of uncertainty width in cases of large bias and extreme asymmetry. The results for the hybrid models support the effectiveness of uncertainty width. They reveal that the Gaussian Process model has the best overall prediction accuracy. The remaining models exhibit diverse trade-offs between precision and accuracy, indicating that model selection should be guided by the specific accuracy and precision requirements for the application.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109474"},"PeriodicalIF":3.9,"publicationDate":"2025-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145517221","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Corrigendum to “Behavioral strategies evolution of stakeholders for wastewater recycling in eco-industrial parks under financial constraints” [Computers & Chemical Engineering, 2025, 204: 109402] “资金约束下生态工业园区废水循环利用的利益相关者行为策略演变”[j] .计算机与化学工程，2025,04:109402。

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-29 DOI: 10.1016/j.compchemeng.2025.109460

Kaixuan Zhang , Xu Han

引用次数: 0

Learning to control inexact Benders decomposition via reinforcement learning 学习通过强化学习控制不精确的bender分解

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-28 DOI: 10.1016/j.compchemeng.2025.109461

Zhe Li , Bernard T. Agyeman , Ilias Mitrai , Prodromos Daoutidis

Benders decomposition (BD), along with its generalized version (GBD), is a widely used algorithm for solving large-scale mixed-integer optimization problems that arise in the operation of process systems. However, the off-the-shelf application to online settings can be computationally inefficient due to the repeated solution of the master problem. An approach to reduce the solution time is to solve the master problem to local optimality. However, identifying the level of suboptimality at each iteration that minimizes the total solution time is nontrivial. In this paper, we propose the application of reinforcement learning to determine the best optimality gap at each GBD iteration. First, we show that the inexact GBD can converge to the optimal solution given a properly designed optimality gap schedule. Next, leveraging reinforcement learning, we learn a policy that minimizes the total solution time, balancing the solution time per iteration with optimality gap improvement. In the resulting RL-iGBD algorithm, the policy adapts the optimality gap at each iteration based on the features of the problem and the solution progress. In numerical experiments on a mixed-integer economic model predictive control problem, we show that the proposed RL-enhanced iGBD method achieves substantial reductions in solution time.

Benders分解（BD）及其广义版本（GBD）是一种广泛应用的算法，用于解决过程系统运行中出现的大规模混合整数优化问题。然而，现成的在线设置应用程序由于主问题的重复解决可能在计算上效率低下。减少求解时间的一种方法是将主问题求解为局部最优。然而，在每次迭代中确定最小化总解决时间的次优性级别是非常重要的。在本文中，我们提出应用强化学习来确定每次GBD迭代的最佳最优性间隙。首先，我们证明了给定一个适当设计的最优性间隙计划，不精确的GBD可以收敛到最优解。接下来，利用强化学习，我们学习一个最小化总解决时间的策略，平衡每次迭代的解决时间和最优性差距改进。在得到的RL-iGBD算法中，策略根据问题的特征和求解进度调整每次迭代的最优性差距。在一个混合整数经济模型预测控制问题的数值实验中，我们证明了所提出的rl增强iGBD方法在求解时间上有很大的缩短。

{"title":"Learning to control inexact Benders decomposition via reinforcement learning","authors":"Zhe Li , Bernard T. Agyeman , Ilias Mitrai , Prodromos Daoutidis","doi":"10.1016/j.compchemeng.2025.109461","DOIUrl":"10.1016/j.compchemeng.2025.109461","url":null,"abstract":"<div><div>Benders decomposition (BD), along with its generalized version (GBD), is a widely used algorithm for solving large-scale mixed-integer optimization problems that arise in the operation of process systems. However, the off-the-shelf application to online settings can be computationally inefficient due to the repeated solution of the master problem. An approach to reduce the solution time is to solve the master problem to local optimality. However, identifying the level of suboptimality at each iteration that minimizes the total solution time is nontrivial. In this paper, we propose the application of reinforcement learning to determine the best optimality gap at each GBD iteration. First, we show that the inexact GBD can converge to the optimal solution given a properly designed optimality gap schedule. Next, leveraging reinforcement learning, we learn a policy that minimizes the total solution time, balancing the solution time per iteration with optimality gap improvement. In the resulting RL-iGBD algorithm, the policy adapts the optimality gap at each iteration based on the features of the problem and the solution progress. In numerical experiments on a mixed-integer economic model predictive control problem, we show that the proposed RL-enhanced iGBD method achieves substantial reductions in solution time.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109461"},"PeriodicalIF":3.9,"publicationDate":"2025-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413890","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A hybrid kernel-based nonparametric system identification approach for multiphase batch processes 基于混合核的多相批处理非参数系统辨识方法

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-27 DOI: 10.1016/j.compchemeng.2025.109472

Shuyu Wang, Zuhua Xu, Jun Zhao, Chunyue Song

In this study, a hybrid kernel-based nonparametric identification approach for multiphase batch processes is proposed. In accordance with the kernel-based identification framework, the time-varying impulse response is modeled as realizations of a zero-mean Gaussian process, whose characteristics are described by the kernel function. However, existing kernel functions for identification are single kernels that cannot accurately describe the abrupt changes in dynamic behaviors caused by multiphase operations. To overcome this limitation, the proposed approach uses a hybrid kernel defined as a set of piecewise kernel functions over partitioned regions to describe different impulse response characteristics. Then, on the basis of the repetitive nature of batch processes, a transition point-based partitioned region representation is developed; it can automatically form a complete/nonoverlapping partition of 2D time–response plane. Building on the partition, we transform the nonparametric identification task into a joint estimation problem for the transition points and kernel hyperparameters, thus overcoming the suboptimality resulting from sequential estimation and improving identification accuracy. Given the introduction of unobservable transition points, we solve the estimation by maximizing the marginal likelihood with the expectation–maximization algorithm. Two cases are studied to demonstrate the effectiveness of the proposed method.

本文提出了一种基于混合核的多相批处理非参数辨识方法。根据基于核的辨识框架，将时变脉冲响应建模为零均值高斯过程的实现，其特征由核函数描述。然而，现有的辨识核函数都是单核函数，无法准确描述多相操作引起的动态行为突变。为了克服这一限制，提出的方法使用混合核定义为一组分段核函数在划分的区域来描述不同的脉冲响应特性。然后，根据批处理过程的重复性，提出了一种基于过渡点的分区区域表示方法；可自动形成二维时间响应平面的完整/不重叠分区。在此基础上，将非参数辨识问题转化为过渡点和核超参数的联合估计问题，克服了序列估计带来的次优性，提高了辨识精度。在引入不可观测过渡点的情况下，我们使用期望最大化算法通过最大化边际似然来解决估计问题。通过两个实例验证了该方法的有效性。

{"title":"A hybrid kernel-based nonparametric system identification approach for multiphase batch processes","authors":"Shuyu Wang, Zuhua Xu, Jun Zhao, Chunyue Song","doi":"10.1016/j.compchemeng.2025.109472","DOIUrl":"10.1016/j.compchemeng.2025.109472","url":null,"abstract":"<div><div>In this study, a hybrid kernel-based nonparametric identification approach for multiphase batch processes is proposed. In accordance with the kernel-based identification framework, the time-varying impulse response is modeled as realizations of a zero-mean Gaussian process, whose characteristics are described by the kernel function. However, existing kernel functions for identification are single kernels that cannot accurately describe the abrupt changes in dynamic behaviors caused by multiphase operations. To overcome this limitation, the proposed approach uses a hybrid kernel defined as a set of piecewise kernel functions over partitioned regions to describe different impulse response characteristics. Then, on the basis of the repetitive nature of batch processes, a transition point-based partitioned region representation is developed; it can automatically form a complete/nonoverlapping partition of 2D time–response plane. Building on the partition, we transform the nonparametric identification task into a joint estimation problem for the transition points and kernel hyperparameters, thus overcoming the suboptimality resulting from sequential estimation and improving identification accuracy. Given the introduction of unobservable transition points, we solve the estimation by maximizing the marginal likelihood with the expectation–maximization algorithm. Two cases are studied to demonstrate the effectiveness of the proposed method.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109472"},"PeriodicalIF":3.9,"publicationDate":"2025-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145463213","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

State-of-health prediction of lithium-ion batteries: A novel PLO-transformer-LSTM framework with DRT feature engineering 锂离子电池健康状态预测：基于DRT特征工程的新型plo -变压器- lstm框架

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-25 DOI: 10.1016/j.compchemeng.2025.109473

Hanfei Wang , Kena Chen , Zhiyu Chen , Siquan Li , Jinjie Wang , Ping Wang , Lijun Yang

Accurate prediction of lithium-ion battery state of health (SOH) is crucial for enhancing the safety and reliability of energy storage systems. However, traditional methods based on capacity decay curves or electrochemical impedance spectroscopy (EIS) often lack interpretability when confronted with complex degradation mechanisms, making it difficult to reveal the specific contributions of different electrochemical processes to capacity decline. This study converts EIS data into a distribution of relaxation times (DRT) format and extracts 12 degradation features from four typical regions as modeling inputs. Building upon this foundation, we propose a predictive framework integrating Transformer with long short-term memory (LSTM) networks. We further introduce the polar light optimization (PLO) algorithm to adaptively optimize three key hyperparameters: attention heads, learning rate, and regularization parameters, thereby enhancing model convergence and predictive capability. Experimental results on the Nature Communications public dataset demonstrate that this method significantly outperforms comparison models such as BP, CNN-LSTM, and Transformer-LSTM under both experimental configurations. In summary, the proposed PLO-Transformer-LSTM method combines high accuracy, robustness, and interpretability, providing a viable pathway for online applications in battery management systems (BMS).

锂离子电池健康状态（SOH）的准确预测对于提高储能系统的安全性和可靠性至关重要。然而，传统的基于容量衰减曲线或电化学阻抗谱（EIS）的方法在面对复杂的降解机制时往往缺乏可解释性，难以揭示不同电化学过程对容量下降的具体贡献。本研究将EIS数据转换为松弛时间分布（DRT）格式，并从4个典型区域提取12个退化特征作为建模输入。在此基础上，我们提出了一个集成Transformer和长短期记忆（LSTM）网络的预测框架。我们进一步引入极光优化（PLO）算法自适应优化三个关键超参数：注意头、学习率和正则化参数，从而提高模型的收敛性和预测能力。在Nature Communications公共数据集上的实验结果表明，该方法在两种实验配置下都明显优于BP、CNN-LSTM和Transformer-LSTM等比较模型。总之，所提出的PLO-Transformer-LSTM方法结合了高精度、鲁棒性和可解释性，为电池管理系统（BMS）的在线应用提供了可行的途径。

{"title":"State-of-health prediction of lithium-ion batteries: A novel PLO-transformer-LSTM framework with DRT feature engineering","authors":"Hanfei Wang , Kena Chen , Zhiyu Chen , Siquan Li , Jinjie Wang , Ping Wang , Lijun Yang","doi":"10.1016/j.compchemeng.2025.109473","DOIUrl":"10.1016/j.compchemeng.2025.109473","url":null,"abstract":"<div><div>Accurate prediction of lithium-ion battery state of health (SOH) is crucial for enhancing the safety and reliability of energy storage systems. However, traditional methods based on capacity decay curves or electrochemical impedance spectroscopy (EIS) often lack interpretability when confronted with complex degradation mechanisms, making it difficult to reveal the specific contributions of different electrochemical processes to capacity decline. This study converts EIS data into a distribution of relaxation times (DRT) format and extracts 12 degradation features from four typical regions as modeling inputs. Building upon this foundation, we propose a predictive framework integrating Transformer with long short-term memory (LSTM) networks. We further introduce the polar light optimization (PLO) algorithm to adaptively optimize three key hyperparameters: attention heads, learning rate, and regularization parameters, thereby enhancing model convergence and predictive capability. Experimental results on the Nature Communications public dataset demonstrate that this method significantly outperforms comparison models such as BP, CNN-LSTM, and Transformer-LSTM under both experimental configurations. In summary, the proposed PLO-Transformer-LSTM method combines high accuracy, robustness, and interpretability, providing a viable pathway for online applications in battery management systems (BMS).</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109473"},"PeriodicalIF":3.9,"publicationDate":"2025-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413897","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Performance degradation prediction of PEMFC based on harris hawks optimization and bidirectional gated recurrent units 基于harris hawks优化和双向门控循环单元的PEMFC性能退化预测

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-25 DOI: 10.1016/j.compchemeng.2025.109466

Tingjie Ba , Guisheng Chen , Qiang Liu , Junwei Yang , Liang Chen , Yinggang Shen , Renxin Xiao , Nan Pan

Proton exchange membrane fuel cell (PEMFC), as highly efficient energy conversion devices, have attracted widespread attention. However, their commercial application is hindered by the issue of performance degradation. This study proposes an innovative data-driven model that integrates the harris hawks optimization (HHO) algorithm and bidirectional gated recurrent units (BiGRU) to accurately predict the performance degradation of PEMFC. BiGRU, an efficient model for processing sequential data, excels at capturing both forward and backward dependencies in time series data, thereby enhancing the accuracy and generalizability of predictions. By employing the HHO algorithm for intelligent optimization of the BiGRU hyperparameters, the HHO-BiGRU model constructed in this paper further improves predictive performance. To verify the effectiveness of the model, four different durability test datasets are used for evaluation. The results show that, with a training set length of 600 h, the HHO-BiGRU model achieves root mean square error (RMSE) values of 0.00031, 0.00011, 0.00031, and 0.00020 on the four datasets, respectively. Especially noteworthy is the model's ability to maintain high accuracy even with minimized training dataset size, significantly lowering the dependence on large datasets and effectively saving costs.

质子交换膜燃料电池（PEMFC）作为一种高效的能量转换装置，受到了广泛的关注。然而，它们的商业应用受到性能下降问题的阻碍。本研究提出了一种创新的数据驱动模型，该模型集成了哈里斯鹰优化（HHO）算法和双向门控循环单元（BiGRU），以准确预测PEMFC的性能下降。BiGRU是一种高效的序列数据处理模型，擅长捕捉时间序列数据的前向和后向依赖关系，从而提高预测的准确性和泛化性。通过采用HHO算法对BiGRU超参数进行智能优化，本文构建的HHO-BiGRU模型进一步提高了预测性能。为了验证模型的有效性，使用了四种不同的耐久性测试数据集进行评估。结果表明，当训练集长度为600 h时，HHO-BiGRU模型在4个数据集上的均方根误差（RMSE）分别为0.00031、0.00011、0.00031和0.00020。特别值得注意的是，该模型即使在最小的训练数据集规模下也能保持较高的准确性，大大降低了对大型数据集的依赖，有效地节省了成本。

{"title":"Performance degradation prediction of PEMFC based on harris hawks optimization and bidirectional gated recurrent units","authors":"Tingjie Ba , Guisheng Chen , Qiang Liu , Junwei Yang , Liang Chen , Yinggang Shen , Renxin Xiao , Nan Pan","doi":"10.1016/j.compchemeng.2025.109466","DOIUrl":"10.1016/j.compchemeng.2025.109466","url":null,"abstract":"<div><div>Proton exchange membrane fuel cell (PEMFC), as highly efficient energy conversion devices, have attracted widespread attention. However, their commercial application is hindered by the issue of performance degradation. This study proposes an innovative data-driven model that integrates the harris hawks optimization (HHO) algorithm and bidirectional gated recurrent units (BiGRU) to accurately predict the performance degradation of PEMFC. BiGRU, an efficient model for processing sequential data, excels at capturing both forward and backward dependencies in time series data, thereby enhancing the accuracy and generalizability of predictions. By employing the HHO algorithm for intelligent optimization of the BiGRU hyperparameters, the HHO-BiGRU model constructed in this paper further improves predictive performance. To verify the effectiveness of the model, four different durability test datasets are used for evaluation. The results show that, with a training set length of 600 h, the HHO-BiGRU model achieves root mean square error (RMSE) values of 0.00031, 0.00011, 0.00031, and 0.00020 on the four datasets, respectively. Especially noteworthy is the model's ability to maintain high accuracy even with minimized training dataset size, significantly lowering the dependence on large datasets and effectively saving costs.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109466"},"PeriodicalIF":3.9,"publicationDate":"2025-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Sustainable optimization of CHOSYN networks in eco-industrial systems via a hybrid machine learning and mathematical programming approach 通过混合机器学习和数学规划方法实现生态工业系统中CHOSYN网络的可持续优化

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-24 DOI: 10.1016/j.compchemeng.2025.109470

Hugo Eduardo Medrano-Minet , Francisco Javier López-Flores , Fabricio Nápoles-Rivera , Eusiel Rubio-Castro , José María Ponce-Ortega

This study presents a hybrid machine learning and mathematical programming framework for the sustainable synthesis of Carbon-Hydrogen-Oxygen Symbiosis Networks (CHOSYN), which promote the internal reuse of hydrocarbon-rich residuals across process units to enhance resource efficiency and reduce environmental impact in eco-industrial systems. A superstructure-based formulation is developed to represent all feasible interconnections between processing units, enabling a systematic exploration of network configurations. Artificial neural networks (ANN), trained on process simulation data, are employed as surrogate models to approximate the operational behavior of complex unit operations. These surrogate models are seamlessly embedded within a superstructure-based optimization framework implemented via the Optimization and Machine Learning Toolkit (OMLT) and solved using Pyomo. A multi-objective optimization approach balances economic profitability and environmental performance. This is the first study on CHOSYN networks in which reaction sections are modeled using machine learning techniques to be coupled with a mathematical optimization model and explicitly optimize internal recirculation, chemical species reuse, and operating conditions within the CHOSYN network, yielding an optimal configuration that demonstrates extensive inter-unit integration. Numerous material exchanges enable the internal reuse of hydrogen, light hydrocarbons, carbon oxides, water, and other valuable streams. Analysis further indicates that high-margin products such as dimethyl ether and propylene contribute significantly to profitability despite lower production volumes, highlighting strategic product selection in integrated networks. Overall, this study establishes a robust and versatile methodology for integrating data-driven models into process network design, offering a promising path toward greener and economically viable chemical manufacturing.

本研究提出了一个用于可持续合成碳-氢-氧共生网络（CHOSYN）的混合机器学习和数学规划框架，该框架促进了整个过程单元中富含碳氢化合物的残留物的内部重复利用，以提高资源效率并减少生态工业系统中的环境影响。开发了基于上层建筑的公式来表示处理单元之间所有可行的互连，从而能够系统地探索网络配置。利用过程仿真数据训练的人工神经网络（ANN）作为替代模型来近似复杂单元操作的操作行为。这些代理模型无缝嵌入到基于上层建筑的优化框架中，该框架通过优化和机器学习工具包（OMLT）实现，并使用Pyomo解决。多目标优化方法平衡了经济效益和环境绩效。这是对CHOSYN网络的第一个研究，其中使用机器学习技术对反应部分进行建模，并将其与数学优化模型相结合，明确优化CHOSYN网络内的内部再循环、化学物质再利用和操作条件，从而产生一个优化配置，展示了广泛的单元间集成。大量的物质交换使氢气、轻烃、碳氧化物、水和其他有价值的流能够在内部重复利用。进一步分析表明，尽管产量下降，但二甲醚和丙烯等高利润产品对盈利能力的贡献很大，突出了综合网络中的战略性产品选择。总体而言，本研究建立了一种强大而通用的方法，将数据驱动模型集成到过程网络设计中，为实现更环保、经济上可行的化学制造提供了一条有希望的道路。

{"title":"Sustainable optimization of CHOSYN networks in eco-industrial systems via a hybrid machine learning and mathematical programming approach","authors":"Hugo Eduardo Medrano-Minet , Francisco Javier López-Flores , Fabricio Nápoles-Rivera , Eusiel Rubio-Castro , José María Ponce-Ortega","doi":"10.1016/j.compchemeng.2025.109470","DOIUrl":"10.1016/j.compchemeng.2025.109470","url":null,"abstract":"<div><div>This study presents a hybrid machine learning and mathematical programming framework for the sustainable synthesis of Carbon-Hydrogen-Oxygen Symbiosis Networks (CHOSYN), which promote the internal reuse of hydrocarbon-rich residuals across process units to enhance resource efficiency and reduce environmental impact in eco-industrial systems. A superstructure-based formulation is developed to represent all feasible interconnections between processing units, enabling a systematic exploration of network configurations. Artificial neural networks (ANN), trained on process simulation data, are employed as surrogate models to approximate the operational behavior of complex unit operations. These surrogate models are seamlessly embedded within a superstructure-based optimization framework implemented via the Optimization and Machine Learning Toolkit (OMLT) and solved using Pyomo. A multi-objective optimization approach balances economic profitability and environmental performance. This is the first study on CHOSYN networks in which reaction sections are modeled using machine learning techniques to be coupled with a mathematical optimization model and explicitly optimize internal recirculation, chemical species reuse, and operating conditions within the CHOSYN network, yielding an optimal configuration that demonstrates extensive inter-unit integration. Numerous material exchanges enable the internal reuse of hydrogen, light hydrocarbons, carbon oxides, water, and other valuable streams. Analysis further indicates that high-margin products such as dimethyl ether and propylene contribute significantly to profitability despite lower production volumes, highlighting strategic product selection in integrated networks. Overall, this study establishes a robust and versatile methodology for integrating data-driven models into process network design, offering a promising path toward greener and economically viable chemical manufacturing.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109470"},"PeriodicalIF":3.9,"publicationDate":"2025-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413891","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A scale-variable attention transformer for industrial process fault diagnosis 一种用于工业过程故障诊断的尺度可变注意力变压器

IF 3.9 2区工程技术 Q2 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers & Chemical Engineering

Pub Date : 2025-10-24 DOI: 10.1016/j.compchemeng.2025.109471

Weidong Chen , Lubin Ye , Xiaofen Fang , Hongfei Guo , Jianqing Li

When implementing the Fault detection and diagnosis (FDD) for industrial process systems, one has to consider the dynamic characterizations. The transformer, which is an attention-based encoder-decoder model, has exhibited outstanding performances in dynamic feature extraction and modeling, and has been validated effective in the natural language processing (NLP) and computer vision. The current transformer-based FDD approaches mainly utilize the time-domain information and have insufficient attention to those oscillatory characterizations. The current work proposes a novel scale-variable attention transformer model, which can extract the time-frequency features in multiple scales to improve the FDD performance. First the continuous wavelet transform (CWT) is adopted to obtain multi-scale expressions of the process variables. Then, a scale-variable attention mechanism is designed to depict the long-time relations on multiple frequency scales and variables. Compared with the tradition transformers, this novel mechanism can obtain attention for each variable and each scale for the multi-scale inputs and thus extract more discriminant dynamic features to help improve the operation condition classification task. The proposed methodology is experimented on the Tennessee Eastman process benchmark for FDD tasks. The testing results demonstrated that the averages of fault detection rate and classification accuracy can reach to about 95 %, which have improvements of 5–15 % compared to existing transformer-based approaches.

在实现工业过程系统的故障检测与诊断（FDD）时，必须考虑动态表征。变压器是一种基于注意力的编码器-解码器模型，在动态特征提取和建模方面表现优异，并在自然语言处理和计算机视觉中得到了有效的验证。基于电流互感器的FDD方法主要利用时域信息，对振荡特性的关注不够。本文提出了一种新颖的尺度可变注意力转换器模型，该模型可以提取多尺度的时频特征，以提高FDD的性能。首先采用连续小波变换（CWT）得到过程变量的多尺度表达式；然后，设计了一个尺度-变量注意机制来描述多个频率尺度和变量之间的长期关系。与传统的变压器相比，该机构可以对多尺度输入的每个变量和每个尺度进行关注，从而提取出更多具有判别性的动态特征，有助于改进运行状态分类任务。提出的方法在FDD任务的Tennessee Eastman过程基准上进行了实验。测试结果表明，该方法的故障检出率和分类准确率平均可达95%左右，与现有的基于变压器的方法相比，提高了5 - 15%。

{"title":"A scale-variable attention transformer for industrial process fault diagnosis","authors":"Weidong Chen , Lubin Ye , Xiaofen Fang , Hongfei Guo , Jianqing Li","doi":"10.1016/j.compchemeng.2025.109471","DOIUrl":"10.1016/j.compchemeng.2025.109471","url":null,"abstract":"<div><div>When implementing the Fault detection and diagnosis (FDD) for industrial process systems, one has to consider the dynamic characterizations. The transformer, which is an attention-based encoder-decoder model, has exhibited outstanding performances in dynamic feature extraction and modeling, and has been validated effective in the natural language processing (NLP) and computer vision. The current transformer-based FDD approaches mainly utilize the time-domain information and have insufficient attention to those oscillatory characterizations. The current work proposes a novel scale-variable attention transformer model, which can extract the time-frequency features in multiple scales to improve the FDD performance. First the continuous wavelet transform (CWT) is adopted to obtain multi-scale expressions of the process variables. Then, a scale-variable attention mechanism is designed to depict the long-time relations on multiple frequency scales and variables. Compared with the tradition transformers, this novel mechanism can obtain attention for each variable and each scale for the multi-scale inputs and thus extract more discriminant dynamic features to help improve the operation condition classification task. The proposed methodology is experimented on the Tennessee Eastman process benchmark for FDD tasks. The testing results demonstrated that the averages of fault detection rate and classification accuracy can reach to about 95 %, which have improvements of 5–15 % compared to existing transformer-based approaches.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109471"},"PeriodicalIF":3.9,"publicationDate":"2025-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413900","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0