Computers & Chemical Engineering最新文献_第9页

Dynamic optimization of water management in aquatic–agricultural systems for net-zero carbon 实现净零碳的水产农业系统水管理动态优化

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.2025.109468

Amira Siniscalchi, Erica P. Schulz, María Soledad Diaz, Guillermo Durand

In this work, we present a novel mathematical model within a dynamic optimization framework to address the management of a salt lake and its basin within a semiarid region, facing extreme weather events. The objective is to propose strategies to mitigate the effects of droughts on lake salinity and agricultural and livestock activities, focusing on achieving the net zero carbon goal at the end of the analyzed time horizon. The model is formulated as an optimal control problem subject to a Differential Algebraic Equations (DAE) system.

We present a case study on Chasicó Lake, a salt lake in a semiarid endorheic basin, in Argentina. The study comprises a five-year drought period that requires freshwater derivation from a constructed reservoir to keep the salt-lake’s salinity within acceptable ranges for reproduction of a valuable fish species. At the same time, sustainable land use and livestock activities must be ensured. Freshwater is used for irrigation of crops, pasture and native species tree as well as for livestock drinking requirements. The control variable is freshwater diverted from the reservoir, and the decision variable is the land area allocated to crops so as to offset livestock emissions. Numerical results demonstrate that net-zero carbon emissions could be achieved while effectively managing drought periods and preserving ecological balance.

在这项工作中，我们在动态优化框架内提出了一个新的数学模型，以解决半干旱地区面临极端天气事件的盐湖及其盆地的管理问题。目标是提出减轻干旱对湖泊盐度以及农业和牲畜活动影响的战略，重点是在分析的时间范围结束时实现净零碳目标。该模型被表述为微分代数方程（DAE）系统的最优控制问题。我们提出了一个案例研究Chasicó湖，一个盐湖在半干旱内陆盆地，在阿根廷。这项研究包括一个为期五年的干旱期，需要从一个建成的水库中提取淡水，以使盐湖的盐度保持在可接受的范围内，以供一种有价值的鱼类繁殖。与此同时，必须确保可持续的土地利用和畜牧业活动。淡水用于灌溉作物、牧场和本地树种以及牲畜的饮水需求。控制变量是水库引水，决策变量是为抵消牲畜排放而分配给作物的土地面积。数值结果表明，在有效管理干旱期和保持生态平衡的同时，可以实现净零碳排放。

{"title":"Dynamic optimization of water management in aquatic–agricultural systems for net-zero carbon","authors":"Amira Siniscalchi, Erica P. Schulz, María Soledad Diaz, Guillermo Durand","doi":"10.1016/j.compchemeng.2025.109468","DOIUrl":"10.1016/j.compchemeng.2025.109468","url":null,"abstract":"<div><div>In this work, we present a novel mathematical model within a dynamic optimization framework to address the management of a salt lake and its basin within a semiarid region, facing extreme weather events. The objective is to propose strategies to mitigate the effects of droughts on lake salinity and agricultural and livestock activities, focusing on achieving the net zero carbon goal at the end of the analyzed time horizon. The model is formulated as an optimal control problem subject to a Differential Algebraic Equations (DAE) system.</div><div>We present a case study on Chasicó Lake, a salt lake in a semiarid endorheic basin, in Argentina. The study comprises a five-year drought period that requires freshwater derivation from a constructed reservoir to keep the salt-lake’s salinity within acceptable ranges for reproduction of a valuable fish species. At the same time, sustainable land use and livestock activities must be ensured. Freshwater is used for irrigation of crops, pasture and native species tree as well as for livestock drinking requirements. The control variable is freshwater diverted from the reservoir, and the decision variable is the land area allocated to crops so as to offset livestock emissions. Numerical results demonstrate that net-zero carbon emissions could be achieved while effectively managing drought periods and preserving ecological balance.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109468"},"PeriodicalIF":3.9,"publicationDate":"2025-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145463214","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

Special issue editorial: Selected contributions from foundations of computer aided process design (FOCAPD) 2024 特刊社论：计算机辅助工艺设计基金会（FOCAPD） 2024

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

Computers & Chemical Engineering

Pub Date : 2025-10-23 DOI: 10.1016/j.compchemeng.2025.109467

Matt Bassett , Thomas A. Adams II , Selen Cremaschi , Monica Zanfir

This short editorial discusses a recent special virtual issue containing papers invited from the best contributions to the FOCAPD 2024 conference. We provide a brief description of each paper and its main contributions to the field. We observe that certain research topics are trending in the field, particularly plastic recycling processes, integrating sustainability metrics and circular economy concepts into process and product design, and process design under uncertainty approaches.

这篇简短的社论讨论了最近的一个特殊的虚拟问题，其中包含了来自FOCAPD 2024年会议的最佳贡献者的论文。我们提供了每篇论文的简要描述及其对该领域的主要贡献。我们观察到该领域的某些研究主题是趋势，特别是塑料回收过程，将可持续性指标和循环经济概念融入过程和产品设计，以及不确定方法下的过程设计。

引用次数: 0

Interpretable online scheduling for chemical batch plants with attention augmented reinforcement learning agents 基于注意力增强强化学习代理的化工批处理厂可解释在线调度

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

Computers & Chemical Engineering

Pub Date : 2025-10-23 DOI: 10.1016/j.compchemeng.2025.109469

Daniel Rangel-Martinez, Luis A. Ricardez-Sandoval

A method for training Deep Reinforcement Learning agents with self-attention modules is presented to address online scheduling decisions in batch chemical plants. The agent is designed to generate multiple decisions at the same time in a partially observable environment; these decisions can be either discrete or continuous. The use of self-attention modules is justified by their increasing popularity in Natural Language Processing (NLP) methods over Recurrent Neural Networks (RNNs). This method leverages the attention-based models by using the attention matrices generated by the agent to produce a numeric interpretation of the agent’s logic. Additionally, a second attention matrix with fixed parameters was embedded in the architecture to define the relevance of each part of the environment. The information from these matrices is demonstrated to be useful for interpreting the relevance of key aspects in the environment and to track the change of the agent’s focal points in non-stationary environments. The methodology is tested using two batch chemical plant case studies, demonstrating the advantages and limitations in comparison to an agent built with RNNs. The interpretability of the decisions and the logic of the agent showed an option to approach the so called black-box characteristic of data-driven models.

提出了一种训练具有自关注模块的深度强化学习智能体的方法，以解决间歇化工厂的在线调度决策问题。该智能体被设计为在部分可观察的环境中同时生成多个决策；这些决策可以是离散的，也可以是连续的。自关注模块的使用在自然语言处理（NLP）方法中越来越受欢迎，而不是在循环神经网络（rnn）中。该方法利用基于注意力的模型，通过使用智能体生成的注意力矩阵来生成智能体逻辑的数值解释。此外，在体系结构中嵌入了具有固定参数的第二个注意力矩阵，以定义环境中每个部分的相关性。来自这些矩阵的信息被证明对于解释环境中关键方面的相关性以及在非固定环境中跟踪代理焦点的变化是有用的。该方法通过两个批量化工厂的案例研究进行了测试，展示了与rnn构建的代理相比的优势和局限性。决策的可解释性和代理的逻辑显示了接近所谓的数据驱动模型的黑箱特征的选项。

{"title":"Interpretable online scheduling for chemical batch plants with attention augmented reinforcement learning agents","authors":"Daniel Rangel-Martinez, Luis A. Ricardez-Sandoval","doi":"10.1016/j.compchemeng.2025.109469","DOIUrl":"10.1016/j.compchemeng.2025.109469","url":null,"abstract":"<div><div>A method for training Deep Reinforcement Learning agents with self-attention modules is presented to address online scheduling decisions in batch chemical plants. The agent is designed to generate multiple decisions at the same time in a partially observable environment; these decisions can be either discrete or continuous. The use of self-attention modules is justified by their increasing popularity in Natural Language Processing (NLP) methods over Recurrent Neural Networks (RNNs). This method leverages the attention-based models by using the attention matrices generated by the agent to produce a numeric interpretation of the agent’s logic. Additionally, a second attention matrix with fixed parameters was embedded in the architecture to define the relevance of each part of the environment. The information from these matrices is demonstrated to be useful for interpreting the relevance of key aspects in the environment and to track the change of the agent’s focal points in non-stationary environments. The methodology is tested using two batch chemical plant case studies, demonstrating the advantages and limitations in comparison to an agent built with RNNs. The interpretability of the decisions and the logic of the agent showed an option to approach the so called black-box characteristic of data-driven models.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109469"},"PeriodicalIF":3.9,"publicationDate":"2025-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413899","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

Visual process monitoring of biomass conversion reactors using transfer learning and generative AI 基于迁移学习和生成式人工智能的生物质转化反应器视觉过程监测

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

Computers & Chemical Engineering

Pub Date : 2025-10-23 DOI: 10.1016/j.compchemeng.2025.109464

Qiangqiang Mao , Boris Yip , Chuanhao Xu , Sagar Garg , Pengcheng Guo , Yankai Cao

Thermochemical reaction system, particularly in the form of small-scale, low-cost bioreactor, offers a promising solution for efficient biomass-to-bioproduct conversion. The bioreactor potentially revolutionizes biomass conversion and keep rural communities into the loop of biomass-based circular economy. However, its continuous operation remains a significant concern. A primary concern arises from the unexpected reaction condition, which is typically indicated by smoke emanating from the bioreactor. This expected issue can be resolved by minor adjustments to ensure proper reaction environment. In this regard, an automated visual process monitoring for smoke detection is crucial. This can be achieved by developing a convolutional neural network (CNN)-based smoke classifier. However, shuffled video-based smoke classification, where a model trained on video recordings from field experiments is applied to monitor new biomass reaction processes with unseen scenarios, poses great challenges due to limited number of field experiments, particularly given the diversity of field backgrounds. Considering the limited diversity issue, this study explore the potential of generative artificial intelligence (GenAI) to generate virtual training images with smoke and without smoke. With augmented training data, a tailored transfer learning strategy is applied to fine tune the CNN model for smoke detection. To assist field operators in understanding classifier decisions and correcting erroneous predictions, an explainable visual representation is provided by smoke localization heatmaps. The results show that the proposed method significantly improve smoke detection accuracy and prediction reliability, which is essential for the continuous operation of biomass conversion reactors and the success of decarbonization within biomass-based circular economy.

热化学反应系统，特别是小型、低成本的生物反应器形式，为生物质的高效转化提供了一个有前途的解决方案。生物反应器可能会彻底改变生物质转化，并使农村社区进入以生物质为基础的循环经济的循环。然而，它的持续运作仍然是一个重大问题。主要的担忧来自意外的反应条件，这通常由生物反应器发出的烟雾表示。这个预期的问题可以通过微小的调整来解决，以确保适当的反应环境。在这方面，一个自动化的视觉过程监测的烟雾检测是至关重要的。这可以通过开发基于卷积神经网络（CNN）的烟雾分类器来实现。然而，由于现场实验数量有限，特别是考虑到现场背景的多样性，基于视频的烟雾分类（在现场实验视频记录上训练的模型被应用于监测未知场景下的新生物质反应过程）带来了巨大的挑战。考虑到有限的多样性问题，本研究探索了生成式人工智能（GenAI）在生成有烟雾和没有烟雾的虚拟训练图像方面的潜力。在增强训练数据的基础上，应用了一种定制的迁移学习策略来微调CNN烟雾探测模型。为了帮助现场操作员理解分类器的决策和纠正错误的预测，烟雾定位热图提供了一个可解释的视觉表示。结果表明，该方法显著提高了烟感探测精度和预测可靠性，为生物质转化反应器的连续运行和生物质循环经济脱碳的成功实现奠定了基础。

{"title":"Visual process monitoring of biomass conversion reactors using transfer learning and generative AI","authors":"Qiangqiang Mao , Boris Yip , Chuanhao Xu , Sagar Garg , Pengcheng Guo , Yankai Cao","doi":"10.1016/j.compchemeng.2025.109464","DOIUrl":"10.1016/j.compchemeng.2025.109464","url":null,"abstract":"<div><div>Thermochemical reaction system, particularly in the form of small-scale, low-cost bioreactor, offers a promising solution for efficient biomass-to-bioproduct conversion. The bioreactor potentially revolutionizes biomass conversion and keep rural communities into the loop of biomass-based circular economy. However, its continuous operation remains a significant concern. A primary concern arises from the unexpected reaction condition, which is typically indicated by smoke emanating from the bioreactor. This expected issue can be resolved by minor adjustments to ensure proper reaction environment. In this regard, an automated visual process monitoring for smoke detection is crucial. This can be achieved by developing a convolutional neural network (CNN)-based smoke classifier. However, shuffled video-based smoke classification, where a model trained on video recordings from field experiments is applied to monitor new biomass reaction processes with unseen scenarios, poses great challenges due to limited number of field experiments, particularly given the diversity of field backgrounds. Considering the limited diversity issue, this study explore the potential of generative artificial intelligence (GenAI) to generate virtual training images with smoke and without smoke. With augmented training data, a tailored transfer learning strategy is applied to fine tune the CNN model for smoke detection. To assist field operators in understanding classifier decisions and correcting erroneous predictions, an explainable visual representation is provided by smoke localization heatmaps. The results show that the proposed method significantly improve smoke detection accuracy and prediction reliability, which is essential for the continuous operation of biomass conversion reactors and the success of decarbonization within biomass-based circular economy.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109464"},"PeriodicalIF":3.9,"publicationDate":"2025-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145360336","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

Scientific machine learning for modeling industrial-scale Primary Separation Vessel 用于工业规模主分离容器建模的科学机器学习

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

Computers & Chemical Engineering

Pub Date : 2025-10-23 DOI: 10.1016/j.compchemeng.2025.109427

Hossein Mohammadghasemi , Jansen Fajar Soesanto , Abhijeet Singh , Bart Maciszewski , Biao Huang

A scientific machine learning (SciML) method integrates masked neural networks with fundamental physical laws to model the Primary Separation Vessel (PSV) in oil sands processing. This framework addresses limitations in conventional models by discovering a critical underlying relation, namely optimized hindered settling functions, through embedded neural networks, subsequently translated into interpretable mathematical expressions via symbolic regression. The methodology significantly enhances computational efficiency through parallel processing and pseudo-transient solver techniques while eliminating the need for scenario-specific parameter tuning. The re-discovered hindered-settling function successfully captures the fundamental behavior of hindered settling across varied conditions, although parameter estimation relied on a limited dataset. Validation against industrial range benchmarks demonstrates the approach’s effectiveness in reproducing characteristic physical behaviors across different ore grades. This advancement represents a significant step in bridging theoretical and data-driven approaches for complex multiphase systems, with promising potential for extension to integrated systems and improved operational optimization in oil sands processing facilities.

一种科学的机器学习（SciML）方法将掩膜神经网络与基本物理定律相结合，对油砂加工中的主分离船（PSV）进行了建模。该框架解决了传统模型的局限性，通过嵌入式神经网络发现了一个关键的潜在关系，即优化的阻碍沉降函数，随后通过符号回归转化为可解释的数学表达式。该方法通过并行处理和伪瞬态求解技术显著提高了计算效率，同时消除了特定场景参数调优的需要。尽管参数估计依赖于有限的数据集，但重新发现的阻碍沉降函数成功地捕获了不同条件下阻碍沉降的基本行为。对工业范围基准的验证表明，该方法在再现不同矿石品位的特征物理行为方面是有效的。这一进步代表了在复杂多相系统的理论和数据驱动方法之间建立桥梁的重要一步，具有扩展到集成系统和改进油砂处理设施操作优化的潜力。

{"title":"Scientific machine learning for modeling industrial-scale Primary Separation Vessel","authors":"Hossein Mohammadghasemi , Jansen Fajar Soesanto , Abhijeet Singh , Bart Maciszewski , Biao Huang","doi":"10.1016/j.compchemeng.2025.109427","DOIUrl":"10.1016/j.compchemeng.2025.109427","url":null,"abstract":"<div><div>A scientific machine learning (SciML) method integrates masked neural networks with fundamental physical laws to model the Primary Separation Vessel (PSV) in oil sands processing. This framework addresses limitations in conventional models by discovering a critical underlying relation, namely optimized hindered settling functions, through embedded neural networks, subsequently translated into interpretable mathematical expressions via symbolic regression. The methodology significantly enhances computational efficiency through parallel processing and pseudo-transient solver techniques while eliminating the need for scenario-specific parameter tuning. The re-discovered hindered-settling function successfully captures the fundamental behavior of hindered settling across varied conditions, although parameter estimation relied on a limited dataset. Validation against industrial range benchmarks demonstrates the approach’s effectiveness in reproducing characteristic physical behaviors across different ore grades. This advancement represents a significant step in bridging theoretical and data-driven approaches for complex multiphase systems, with promising potential for extension to integrated systems and improved operational optimization in oil sands processing facilities.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109427"},"PeriodicalIF":3.9,"publicationDate":"2025-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413901","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

Novel sufficient conditions for the local stability of non-isothermal continuous homogeneous reaction systems 非等温连续均相反应体系局部稳定性的新充分条件

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

Computers & Chemical Engineering

Pub Date : 2025-10-22 DOI: 10.1016/j.compchemeng.2025.109462

N. Ha Hoang , D. Bonvin

This note investigates the stability of non-isothermal continuous homogeneous reaction systems involving

S

species,

R

reactions,

p

inlet streams, and one outlet stream. The analysis, which is based on Lyapunov’s indirect method, is greatly simplified by transforming the reactor model of

S + 1

dynamic equations obtained from material and energy balances into

R + 1

dynamic equations expressed in terms of vessel extents of reaction and heat exchange. By linearizing this reduced model about the equilibrium point of interest, stability conditions can be established by computing the eigenvalues of a reduced system matrix. Furthermore, for the case of a single reaction of any order that obeys mass-action kinetics, novel sufficient stability conditions have been developed, which do not require computing eigenvalues. These stability conditions are proven to always hold for endothermic reactions using thermodynamical arguments. In addition, it is shown that the proposed stability conditions can be relaxed in the case of exothermic reactions depending on the value of heat-transfer coefficient. Two CSTR examples, with steady-state uniqueness and multiplicity behavior respectively, are used to illustrate the theoretical developments.

本文研究了涉及S种、R个反应、p个入口流和一个出口流的非等温连续均相反应系统的稳定性。该分析基于Lyapunov的间接方法，通过将由物质和能量平衡得到的S+1动力学方程的反应器模型转化为以反应和热交换的容器范围表示的R+1动力学方程，大大简化了分析。通过线性化这个关于平衡兴趣点的简化模型，可以通过计算简化系统矩阵的特征值来建立稳定性条件。此外，对于服从质量作用动力学的任何阶的单一反应，已经开发了新的足够的稳定性条件，它不需要计算特征值。用热力学论据证明了这些稳定性条件对吸热反应总是成立的。此外，还表明，放热反应的稳定性条件可以随传热系数的取值而放宽。用两个分别具有稳态唯一性和多重性的CSTR实例说明了理论的发展。

{"title":"Novel sufficient conditions for the local stability of non-isothermal continuous homogeneous reaction systems","authors":"N. Ha Hoang , D. Bonvin","doi":"10.1016/j.compchemeng.2025.109462","DOIUrl":"10.1016/j.compchemeng.2025.109462","url":null,"abstract":"<div><div>This note investigates the stability of non-isothermal continuous homogeneous reaction systems involving <span><math><mi>S</mi></math></span> species, <span><math><mi>R</mi></math></span> reactions, <span><math><mi>p</mi></math></span> inlet streams, and one outlet stream. The analysis, which is based on Lyapunov’s indirect method, is greatly simplified by transforming the reactor model of <span><math><mrow><mi>S</mi><mo>+</mo><mn>1</mn></mrow></math></span> dynamic equations obtained from material and energy balances into <span><math><mrow><mi>R</mi><mo>+</mo><mn>1</mn></mrow></math></span> dynamic equations expressed in terms of vessel extents of reaction and heat exchange. By linearizing this reduced model about the equilibrium point of interest, stability conditions can be established by computing the eigenvalues of a reduced system matrix. Furthermore, for the case of a single reaction of any order that obeys mass-action kinetics, novel sufficient stability conditions have been developed, which do not require computing eigenvalues. These stability conditions are proven to always hold for endothermic reactions using thermodynamical arguments. In addition, it is shown that the proposed stability conditions can be relaxed in the case of exothermic reactions depending on the value of heat-transfer coefficient. Two CSTR examples, with steady-state uniqueness and multiplicity behavior respectively, are used to illustrate the theoretical developments.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109462"},"PeriodicalIF":3.9,"publicationDate":"2025-10-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413889","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

Early prediction of lithium-ion battery life using a hybrid deep neural network and ensemble learning approach 使用混合深度神经网络和集成学习方法对锂离子电池寿命进行早期预测

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

Computers & Chemical Engineering

Pub Date : 2025-10-22 DOI: 10.1016/j.compchemeng.2025.109465

Chuqing Cao , Han Gao , Fengnan Liu , Muhaned Maysrah , Yixuan Zhou , Lin Chen

Accurate early-stage prediction of the remaining useful life (RUL) of lithium-ion batteries is essential for effective repurposing and second-life utilization, thereby advancing sustainable energy storage. Reliable RUL estimation enhances operational efficiency and safety in applications such as electric vehicles and portable electronics. However, early-cycle prediction is hindered by nonlinear degradation behavior and limited initial data. In this study, we propose a Hybrid Deep Neural Network (HDNN) with an adaptive ensemble learning strategy to address these challenges. The HDNN combines CNN, BiLSTM, and GRU layers as its core architecture. To improve interpretability and reduce redundancy, a hybrid feature selection strategy is employed, integrating Spearman correlation, XGBoost ranking, and domain-informed features. An adaptive ensemble framework is further developed, where individual models are dynamically weighted by inverse RMSE to enhance prediction robustness. Using only the first 100 charge–discharge cycles from a widely recognized dataset, accurate early-life RUL estimation is achieved. Extensive experiments show that the framework outperforms baseline and state-of-the-art models, achieving an RMSE of 91.138, R² of 0.94, and MAPE of 11.21%. These results demonstrate the effectiveness of combining a hybrid neural architecture, hybrid feature selection, and adaptive ensemble learning for robust early-stage RUL prediction. This approach improves battery assessment, supports proactive maintenance, and optimizes energy storage management.

准确的早期预测锂离子电池剩余使用寿命（RUL）对于有效再利用和二次使用至关重要，从而促进可持续能源储存。可靠的RUL估计提高了电动汽车和便携式电子设备等应用的操作效率和安全性。然而，非线性退化行为和有限的初始数据阻碍了早期周期预测。在本研究中，我们提出了一种具有自适应集成学习策略的混合深度神经网络（HDNN）来解决这些挑战。HDNN将CNN、BiLSTM、GRU三层作为核心架构。为了提高可解释性和减少冗余，采用了一种混合特征选择策略，将Spearman相关、XGBoost排序和领域通知特征集成在一起。进一步开发了自适应集成框架，其中单个模型通过逆RMSE动态加权以增强预测鲁棒性。仅使用来自广泛认可的数据集的前100个充放电周期，就可以实现准确的早期寿命RUL估计。大量实验表明，该框架优于基线和最先进的模型，RMSE为91.138，R²为0.94，MAPE为11.21%。这些结果证明了将混合神经结构、混合特征选择和自适应集成学习相结合用于鲁棒早期RUL预测的有效性。该方法改进了电池评估，支持主动维护，并优化了储能管理。

{"title":"Early prediction of lithium-ion battery life using a hybrid deep neural network and ensemble learning approach","authors":"Chuqing Cao , Han Gao , Fengnan Liu , Muhaned Maysrah , Yixuan Zhou , Lin Chen","doi":"10.1016/j.compchemeng.2025.109465","DOIUrl":"10.1016/j.compchemeng.2025.109465","url":null,"abstract":"<div><div>Accurate early-stage prediction of the remaining useful life (RUL) of lithium-ion batteries is essential for effective repurposing and second-life utilization, thereby advancing sustainable energy storage. Reliable RUL estimation enhances operational efficiency and safety in applications such as electric vehicles and portable electronics. However, early-cycle prediction is hindered by nonlinear degradation behavior and limited initial data. In this study, we propose a Hybrid Deep Neural Network (HDNN) with an adaptive ensemble learning strategy to address these challenges. The HDNN combines CNN, BiLSTM, and GRU layers as its core architecture. To improve interpretability and reduce redundancy, a hybrid feature selection strategy is employed, integrating Spearman correlation, XGBoost ranking, and domain-informed features. An adaptive ensemble framework is further developed, where individual models are dynamically weighted by inverse RMSE to enhance prediction robustness. Using only the first 100 charge–discharge cycles from a widely recognized dataset, accurate early-life RUL estimation is achieved. Extensive experiments show that the framework outperforms baseline and state-of-the-art models, achieving an RMSE of 91.138, R² of 0.94, and MAPE of 11.21%. These results demonstrate the effectiveness of combining a hybrid neural architecture, hybrid feature selection, and adaptive ensemble learning for robust early-stage RUL prediction. This approach improves battery assessment, supports proactive maintenance, and optimizes energy storage management.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109465"},"PeriodicalIF":3.9,"publicationDate":"2025-10-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413892","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

Improving selectivity in the butene/isobutane alkylation process 提高丁烯/异丁烷烷基化过程的选择性

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

Computers & Chemical Engineering

Pub Date : 2025-10-21 DOI: 10.1016/j.compchemeng.2025.109463

William L. Luyben

Many chemical reaction systems involve the simultaneous production of desirable and undesirable products. Unless selectivity is large, yield and profitability are adversely affected. Developing a process that balances the many engineering tradeoffs inherent in such a system presents many challenging technical and economic problems. The alkylation process used in petroleum refineries around the world is an important example. Butene and isobutane react to form the desired high-octane-number isooctane product (RO 100) and the undesired very-low-octane-number dodecane (RO –20). The alkylate mixture is blended into gasoline, and higher selectivity gives more gasoline. A large isobutane recycle is used to adjust reactor composition, which impacts selectivity, but more recycle increases energy consumption in the separation section. Another process feature that increases design complexity is the presence of two inert components in two fresh feed streams.

This paper presents a quantitative study of the design of an alkylation process to access the interacting process and economic effects over a wide range of process parameters. Results indicate that economics improve when recycle flowrates are significantly larger than those currently used in existing plants.

许多化学反应系统都涉及需要和不需要的产物同时产生。除非选择性很大，否则产量和盈利能力会受到不利影响。在这样一个系统中，开发一个过程来平衡许多固有的工程权衡，提出了许多具有挑战性的技术和经济问题。世界各地炼油厂使用的烷基化工艺就是一个重要的例子。丁烯和异丁烷反应生成所需的高辛烷值异辛烷产品（RO - 100）和所需的极低辛烷值十二烷（RO -20）。将烷基酸混合物掺入汽油中，选择性越高，汽油产量越高。大的异丁烷循环用于调整反应器组成，影响选择性，但更多的循环增加了分离段的能耗。另一个增加设计复杂性的工艺特征是在两个新鲜进料流中存在两个惰性成分。本文对烷基化工艺的设计进行了定量研究，以获得在广泛的工艺参数范围内的相互作用过程和经济效果。结果表明，当循环流量明显大于现有装置的流量时，经济性得到提高。

{"title":"Improving selectivity in the butene/isobutane alkylation process","authors":"William L. Luyben","doi":"10.1016/j.compchemeng.2025.109463","DOIUrl":"10.1016/j.compchemeng.2025.109463","url":null,"abstract":"<div><div>Many chemical reaction systems involve the simultaneous production of desirable and undesirable products. Unless selectivity is large, yield and profitability are adversely affected. Developing a process that balances the many engineering tradeoffs inherent in such a system presents many challenging technical and economic problems. The alkylation process used in petroleum refineries around the world is an important example. Butene and isobutane react to form the desired high-octane-number isooctane product (RO 100) and the undesired very-low-octane-number dodecane (RO –20). The alkylate mixture is blended into gasoline, and higher selectivity gives more gasoline. A large isobutane recycle is used to adjust reactor composition, which impacts selectivity, but more recycle increases energy consumption in the separation section. Another process feature that increases design complexity is the presence of two inert components in two fresh feed streams.</div><div>This paper presents a quantitative study of the design of an alkylation process to access the interacting process and economic effects over a wide range of process parameters. Results indicate that economics improve when recycle flowrates are significantly larger than those currently used in existing plants.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109463"},"PeriodicalIF":3.9,"publicationDate":"2025-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145413926","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

Pipeline pressure drop modeling and compression power minimization for nonideal, compressible, gas mixture flows with application to pipeline flow of natural gas-hydrogen blends 非理想可压缩气体混合流的管道压降建模和压缩功率最小化及其在天然气-氢混合物管道流中的应用

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

Computers & Chemical Engineering

Pub Date : 2025-10-16 DOI: 10.1016/j.compchemeng.2025.109455

Jeremy Conner, Vasilios I. Manousiouthakis

This work presents a novel, dimensionless model that results in a dimensionless algebraic equation that can be used to quantify the pressure drop associated with the steady state, isothermal flow through a straight, horizontal pipeline, of a compressible gas mixture whose thermodynamic behavior is described for comparison purposes by ideal gas (IG) and nonideal gas generic cubic (GC) equation of state (EOS) models. A solution strategy, that uses the aforementioned dimensionless algebraic equation, is then presented that quantifies the pipeline pressure drop for hydrogen containing mixtures. A novel power minimization formulation for pipeline gas compression is then presented, which utilizes the aforementioned dimensionless pipeline pressure drop model for a fluid with GCEOS thermodynamics. Several case studies are then presented to illustrate the pipeline pressure drop dependence on the hydrogen mole fraction of a binary, methane hydrogen mixture, and a natural gas hydrogen mixture with a real life natural gas composition containing eight species, and the optimum power of compression for a pipeline used to deliver constant power for various methane-hydrogen composition blends.

这项工作提出了一种新颖的无量纲模型，该模型产生了一个无量纲代数方程，可用于量化与稳定状态相关的压降，通过直水平管道的等温流动，可压缩气体混合物的热力学行为通过理想气体（IG）和非理想气体一般立方（GC）状态方程（EOS）模型进行比较。然后提出了一种利用上述无量纲代数方程的求解策略，用于量化含氢混合物的管道压降。然后提出了一种新的管道气体压缩功率最小化公式，该公式利用上述无量纲管道压降模型对具有GCEOS热力学的流体进行压缩。然后提出了几个案例研究，以说明管道压降依赖于二元、甲烷氢混合物和天然气氢混合物的氢摩尔分数，其中实际天然气成分含有八种，以及用于输送各种甲烷氢混合物恒定功率的管道的最佳压缩功率。

{"title":"Pipeline pressure drop modeling and compression power minimization for nonideal, compressible, gas mixture flows with application to pipeline flow of natural gas-hydrogen blends","authors":"Jeremy Conner, Vasilios I. Manousiouthakis","doi":"10.1016/j.compchemeng.2025.109455","DOIUrl":"10.1016/j.compchemeng.2025.109455","url":null,"abstract":"<div><div>This work presents a novel, dimensionless model that results in a dimensionless algebraic equation that can be used to quantify the pressure drop associated with the steady state, isothermal flow through a straight, horizontal pipeline, of a compressible gas mixture whose thermodynamic behavior is described for comparison purposes by ideal gas (IG) and nonideal gas generic cubic (GC) equation of state (EOS) models. A solution strategy, that uses the aforementioned dimensionless algebraic equation, is then presented that quantifies the pipeline pressure drop for hydrogen containing mixtures. A novel power minimization formulation for pipeline gas compression is then presented, which utilizes the aforementioned dimensionless pipeline pressure drop model for a fluid with GCEOS thermodynamics. Several case studies are then presented to illustrate the pipeline pressure drop dependence on the hydrogen mole fraction of a binary, methane hydrogen mixture, and a natural gas hydrogen mixture with a real life natural gas composition containing eight species, and the optimum power of compression for a pipeline used to deliver constant power for various methane-hydrogen composition blends.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109455"},"PeriodicalIF":3.9,"publicationDate":"2025-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145360334","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 by design: A first attempt on bioprocessing 可持续设计：生物处理的第一次尝试

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

Computers & Chemical Engineering

Pub Date : 2025-10-15 DOI: 10.1016/j.compchemeng.2025.109454

Miriam Sarkis , Mariana Monteiro , Andrea Bernardi , Ranjith Chiplunkar , Cleo Kontoravdi , Maria M. Papathanasiou

The growing commitment of the biopharmaceutical sector to transition to Net Zero is driving the industry to embed sustainability principles across its entire pipeline of operations from early-stage process development to manufacturing. In this context, a key challenge for process design is the prediction of the impact of upstream variability on downstream process performance and, therefore, design, with effects on process economics and sustainability. In this work, we focus on the economic and sustainability analysis of antibody-producing bioprocess designs in the presence and absence of downstream process performance constraints. Specifically, we introduce a kinetic model of upstream processing that predicts the profile of critical cell-derived and product-associated impurities and their variability based on culture conditions. Upstream model simulation results are then used to inform a superstructure optimization that maximizes monoclonal antibody (mAb) throughput under purity constraints. Flowsheet simulation models of the candidate designs are developed and process performance is evaluated through techno-economic and life cycle assessment. As expected, results show that purity constraints can lead to more complex downstream configurations, with higher nominal costs and footprint, and improved capacity to withstand feedstock variability. Although intuitive, the results highlight the significance of uncertainty quantification and impurity modeling for informing end-to-end process design. The digitally-enabled holistic approach proposed herein comprehensively enables cost-effective, eco-efficient, and uncertainty-aware design decisions in bioprocessing.

生物制药行业越来越多地致力于向净零排放过渡，这推动了该行业从早期工艺开发到制造的整个运营管道中嵌入可持续性原则。在这种情况下，工艺设计的一个关键挑战是预测上游可变性对下游工艺性能的影响，从而预测设计对工艺经济性和可持续性的影响。在这项工作中，我们专注于在存在和不存在下游工艺性能限制的情况下，抗体生产生物工艺设计的经济和可持续性分析。具体来说，我们介绍了一个上游处理的动力学模型，该模型预测了关键细胞衍生和产品相关杂质的分布及其基于培养条件的可变性。然后使用上游模型仿真结果来告知上层结构优化，在纯度限制下最大化单克隆抗体（mAb）的吞吐量。开发了候选设计的流程仿真模型，并通过技术经济和生命周期评估对工艺性能进行了评估。正如预期的那样，结果表明，纯度限制可能导致更复杂的下游配置，具有更高的名义成本和足迹，并提高了承受原料变化的能力。虽然直观，但结果突出了不确定性量化和杂质建模对通知端到端工艺设计的重要性。本文提出的数字化整体方法全面实现了生物处理中具有成本效益，生态效率和不确定性意识的设计决策。

{"title":"Sustainable by design: A first attempt on bioprocessing","authors":"Miriam Sarkis , Mariana Monteiro , Andrea Bernardi , Ranjith Chiplunkar , Cleo Kontoravdi , Maria M. Papathanasiou","doi":"10.1016/j.compchemeng.2025.109454","DOIUrl":"10.1016/j.compchemeng.2025.109454","url":null,"abstract":"<div><div>The growing commitment of the biopharmaceutical sector to transition to Net Zero is driving the industry to embed sustainability principles across its entire pipeline of operations from early-stage process development to manufacturing. In this context, a key challenge for process design is the prediction of the impact of upstream variability on downstream process performance and, therefore, design, with effects on process economics and sustainability. In this work, we focus on the economic and sustainability analysis of antibody-producing bioprocess designs in the presence and absence of downstream process performance constraints. Specifically, we introduce a kinetic model of upstream processing that predicts the profile of critical cell-derived and product-associated impurities and their variability based on culture conditions. Upstream model simulation results are then used to inform a superstructure optimization that maximizes monoclonal antibody (mAb) throughput under purity constraints. Flowsheet simulation models of the candidate designs are developed and process performance is evaluated through techno-economic and life cycle assessment. As expected, results show that purity constraints can lead to more complex downstream configurations, with higher nominal costs and footprint, and improved capacity to withstand feedstock variability. Although intuitive, the results highlight the significance of uncertainty quantification and impurity modeling for informing end-to-end process design. The digitally-enabled holistic approach proposed herein comprehensively enables cost-effective, eco-efficient, and uncertainty-aware design decisions in bioprocessing.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109454"},"PeriodicalIF":3.9,"publicationDate":"2025-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145360246","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