Pub Date : 2025-11-14DOI: 10.1016/j.jmsy.2025.11.009
Ziling Wang , Lai Zou , Wenxi Wang , A.Y.C. Nee , S.K. Ong
The robotic multi-region grinding method is used to machine curved workpieces with uneven surface allowance to further improve their profile accuracy. However, in this method, when the robot executes toolpaths across adjacent grinding regions, force changes and non-uniform grinding often occur. To address the problem, a novel cross-region robotic grinding method that incorporates force control and toolpath planning across adjacent regions is proposed. In this method, a toolpath planning method, with consideration of multiple grinding regions with different expected grinding forces, is developed to generate cutter-contact (CC) points in each toolpath curve based on the point clouds of workpieces, and ensure that there are CC points near the boundary between two adjacent grinding regions. Furthermore, a novel model predictive control scheme with an environment observer is designed to track the grinding force in a single grinding region. In addition, the adaptive impedance model with a novel adaptive update rate is introduced into the control scheme to reduce the changes in the grinding force along the toolpaths across two adjacent regions. Robotic grinding experiments are conducted to verify the superiority of the proposed grinding method. The surface accuracy of the curved workpiece is improved by some 26 %.
{"title":"Cross-region robotic grinding with adaptive toolpath planning and force control for point clouds of complex curved workpieces","authors":"Ziling Wang , Lai Zou , Wenxi Wang , A.Y.C. Nee , S.K. Ong","doi":"10.1016/j.jmsy.2025.11.009","DOIUrl":"10.1016/j.jmsy.2025.11.009","url":null,"abstract":"<div><div>The robotic multi-region grinding method is used to machine curved workpieces with uneven surface allowance to further improve their profile accuracy. However, in this method, when the robot executes toolpaths across adjacent grinding regions, force changes and non-uniform grinding often occur. To address the problem, a novel cross-region robotic grinding method that incorporates force control and toolpath planning across adjacent regions is proposed. In this method, a toolpath planning method, with consideration of multiple grinding regions with different expected grinding forces, is developed to generate cutter-contact (CC) points in each toolpath curve based on the point clouds of workpieces, and ensure that there are CC points near the boundary between two adjacent grinding regions. Furthermore, a novel model predictive control scheme with an environment observer is designed to track the grinding force in a single grinding region. In addition, the adaptive impedance model with a novel adaptive update rate is introduced into the control scheme to reduce the changes in the grinding force along the toolpaths across two adjacent regions. Robotic grinding experiments are conducted to verify the superiority of the proposed grinding method. The surface accuracy of the curved workpiece is improved by some 26 %.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 856-877"},"PeriodicalIF":14.2,"publicationDate":"2025-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516968","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-12DOI: 10.1016/j.jmsy.2025.11.007
Junhyeong Lee , Joon-Young Kim , Heekyu Kim , Inhyo Lee , Seunghwa Ryu
The injection molding industry faces critical challenges in preserving and transferring field knowledge, particularly as experienced workers retire and multilingual barriers hinder effective communication. This study introduces IM-Chat, a multi-agent framework based on large language models (LLMs), designed to facilitate knowledge transfer in injection molding. IM-Chat integrates both limited documented knowledge (e.g., troubleshooting tables, manuals) and extensive field data modeled through a data-driven process condition generator that infers optimal manufacturing settings from environmental inputs such as temperature and humidity, enabling robust and context-aware task resolution. By adopting a retrieval-augmented generation (RAG) strategy and tool-calling agents within a modular architecture, IM-Chat ensures adaptability without the need for fine-tuning. Performance was assessed across 100 single-tool and 60 hybrid tasks for GPT-4o, GPT-4o-mini, and GPT-3.5-turbo by domain experts using a 10-point rubric focused on relevance and correctness, and was further supplemented by automated evaluation using GPT-4o guided by a domain-adapted instruction prompt. The evaluation results indicate that more capable models tend to achieve higher accuracy, particularly in complex, tool-integrated scenarios. In addition, compared with the fine-tuned single-agent LLM, IM-Chat demonstrated superior accuracy, particularly in quantitative reasoning, and greater scalability in handling multiple information sources. Overall, these findings demonstrate the viability of multi-agent LLM systems for industrial knowledge workflows and establish IM-Chat as a scalable and generalizable approach to AI-assisted decision support in manufacturing. To support reproducibility and practical adoption, supplementary materials including prompts, evaluation data, and video demonstrations are made available.
{"title":"IM-Chat: A multi-agent LLM framework integrating tool-calling and diffusion modeling for knowledge transfer in injection molding industry","authors":"Junhyeong Lee , Joon-Young Kim , Heekyu Kim , Inhyo Lee , Seunghwa Ryu","doi":"10.1016/j.jmsy.2025.11.007","DOIUrl":"10.1016/j.jmsy.2025.11.007","url":null,"abstract":"<div><div>The injection molding industry faces critical challenges in preserving and transferring field knowledge, particularly as experienced workers retire and multilingual barriers hinder effective communication. This study introduces IM-Chat, a multi-agent framework based on large language models (LLMs), designed to facilitate knowledge transfer in injection molding. IM-Chat integrates both limited documented knowledge (e.g., troubleshooting tables, manuals) and extensive field data modeled through a data-driven process condition generator that infers optimal manufacturing settings from environmental inputs such as temperature and humidity, enabling robust and context-aware task resolution. By adopting a retrieval-augmented generation (RAG) strategy and tool-calling agents within a modular architecture, IM-Chat ensures adaptability without the need for fine-tuning. Performance was assessed across 100 single-tool and 60 hybrid tasks for GPT-4o, GPT-4o-mini, and GPT-3.5-turbo by domain experts using a 10-point rubric focused on relevance and correctness, and was further supplemented by automated evaluation using GPT-4o guided by a domain-adapted instruction prompt. The evaluation results indicate that more capable models tend to achieve higher accuracy, particularly in complex, tool-integrated scenarios. In addition, compared with the fine-tuned single-agent LLM, IM-Chat demonstrated superior accuracy, particularly in quantitative reasoning, and greater scalability in handling multiple information sources. Overall, these findings demonstrate the viability of multi-agent LLM systems for industrial knowledge workflows and establish IM-Chat as a scalable and generalizable approach to AI-assisted decision support in manufacturing. To support reproducibility and practical adoption, <span><span>supplementary materials</span></span> including prompts, evaluation data, and video demonstrations are made available.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 839-855"},"PeriodicalIF":14.2,"publicationDate":"2025-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516966","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-12DOI: 10.1016/j.jmsy.2025.10.014
Yanying Wang , Zhiheng Zhao , Yujie Han , Ying Cheng , George Q. Huang
Fast fashion platforms such as SHEIN coordinate thousands of small, nearby garment factories to fulfil large numbers of small-lot, fast-switch orders. Dispatching each order operation to the most suitable factory while respecting process routes creates a Flexible Job Shop Scheduling Problem with sequence-dependent setup times (FJSP-SDST), which is NP-hard and must be solved repeatedly as new batches arrive. Current end-to-end deep reinforcement learning (DRL) schedulers either reschedule every batch from scratch or disregard setup costs, which undermines efficiency. We introduce a Cyber-Physical Internet (CPI) scheduling framework that provides routers for every factory and the company, enabling them to cache solved schedules and broadcast real time factory states. This approach skips redundant computations and supplies fresh setup time data. Within this framework, we have developed a Hierarchical Attention Network based Reinforcement Learning (HANRL) scheduler to model the interactions between orders and factories, as well as factory competition and setup costs. Experiments on synthetic and public benchmarks demonstrate that HANRL reduces makespan and improves generalization over state-of-the-art DRL baselines, all while retaining sub-second decision times. This proves the suitability of HANRL for large scale social manufacturing environments.
{"title":"Cyber-physical internet enabled hierarchical attention network based reinforcement learning for order dispatch in fast fashion manufacturing","authors":"Yanying Wang , Zhiheng Zhao , Yujie Han , Ying Cheng , George Q. Huang","doi":"10.1016/j.jmsy.2025.10.014","DOIUrl":"10.1016/j.jmsy.2025.10.014","url":null,"abstract":"<div><div>Fast fashion platforms such as SHEIN coordinate thousands of small, nearby garment factories to fulfil large numbers of small-lot, fast-switch orders. Dispatching each order operation to the most suitable factory while respecting process routes creates a Flexible Job Shop Scheduling Problem with sequence-dependent setup times (FJSP-SDST), which is NP-hard and must be solved repeatedly as new batches arrive. Current end-to-end deep reinforcement learning (DRL) schedulers either reschedule every batch from scratch or disregard setup costs, which undermines efficiency. We introduce a Cyber-Physical Internet (CPI) scheduling framework that provides routers for every factory and the company, enabling them to cache solved schedules and broadcast real time factory states. This approach skips redundant computations and supplies fresh setup time data. Within this framework, we have developed a Hierarchical Attention Network based Reinforcement Learning (HANRL) scheduler to model the interactions between orders and factories, as well as factory competition and setup costs. Experiments on synthetic and public benchmarks demonstrate that HANRL reduces makespan and improves generalization over state-of-the-art DRL baselines, all while retaining sub-second decision times. This proves the suitability of HANRL for large scale social manufacturing environments.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 784-799"},"PeriodicalIF":14.2,"publicationDate":"2025-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516964","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-12DOI: 10.1016/j.jmsy.2025.11.005
Hongquan Gui , Zhanpeng Yang , Arjun Rachana Harish , Cheng Ren , Yishu Yang , Ming Li
Customized garment production is hindered by the expert-dependent nature of sewing pattern generation—a skill-intensive process requiring years of training. While recent approaches aim to translate user intent into sewing patterns, they often struggle to interpret multimodal inputs such as text and images. Multimodal large language models (MLLMs) offer a promising path forward, as they can naturally understand diverse user intents. Yet, applying MLLMs to sewing pattern generation is challenging because conventional tokenization methods often lose the structural information of sewing patterns. To address this issue, we propose GenPattern, a novel framework that integrates structured graph modeling with MLLMs to enable more accurate sewing pattern generation. We introduce a scalable vector graphics (SVG)-style pattern tokenizer, which encodes sewing patterns into structured token sequences. Furthermore, we present SewGraphFuser, a dual-graph module that explicitly models geometric and semantic dependencies to inject structural information into MLLMs. This module combines a structure graph convolution module and a sequence graph convolution module to jointly capture multi-scale spatial and sequential features via a geometric consistency graph and a semantic dependency graph. Finally, to bridge the gap between digital design and physical fabrication, our framework drives a human-robot collaborative cutting platform, enabling expert-free, on-demand garment customization. This innovation empowers human-robot collaboration in pattern production, enhancing scalability in real-world manufacturing. Experimental results show that GenPattern achieves 86.7 % stitch accuracy and reduces panel vertex L2 error to 2.9 cm, demonstrating its potential to democratize custom fashion by enabling non-experts to reliably produce physical garments directly from their ideas.
{"title":"GenPattern: dual-graph enhanced sewing pattern generation via multimodal large language model","authors":"Hongquan Gui , Zhanpeng Yang , Arjun Rachana Harish , Cheng Ren , Yishu Yang , Ming Li","doi":"10.1016/j.jmsy.2025.11.005","DOIUrl":"10.1016/j.jmsy.2025.11.005","url":null,"abstract":"<div><div>Customized garment production is hindered by the expert-dependent nature of sewing pattern generation—a skill-intensive process requiring years of training. While recent approaches aim to translate user intent into sewing patterns, they often struggle to interpret multimodal inputs such as text and images. Multimodal large language models (MLLMs) offer a promising path forward, as they can naturally understand diverse user intents. Yet, applying MLLMs to sewing pattern generation is challenging because conventional tokenization methods often lose the structural information of sewing patterns. To address this issue, we propose GenPattern, a novel framework that integrates structured graph modeling with MLLMs to enable more accurate sewing pattern generation. We introduce a scalable vector graphics (SVG)-style pattern tokenizer, which encodes sewing patterns into structured token sequences. Furthermore, we present SewGraphFuser, a dual-graph module that explicitly models geometric and semantic dependencies to inject structural information into MLLMs. This module combines a structure graph convolution module and a sequence graph convolution module to jointly capture multi-scale spatial and sequential features via a geometric consistency graph and a semantic dependency graph. Finally, to bridge the gap between digital design and physical fabrication, our framework drives a human-robot collaborative cutting platform, enabling expert-free, on-demand garment customization. This innovation empowers human-robot collaboration in pattern production, enhancing scalability in real-world manufacturing. Experimental results show that GenPattern achieves 86.7 % stitch accuracy and reduces panel vertex L2 error to 2.9 cm, demonstrating its potential to democratize custom fashion by enabling non-experts to reliably produce physical garments directly from their ideas.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 822-838"},"PeriodicalIF":14.2,"publicationDate":"2025-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516965","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-11DOI: 10.1016/j.jmsy.2025.11.010
José Joaquín Peralta Abadía , Fabio Marco Monetti , Sylvia Nathaly Rea Minango , Angela Carrera-Rivera , Miriam Ugarte Querejeta , Mikel Cuesta Zabaljauregui , Felix Larrinaga Barrenechea , Miren Illarramendi Rezabal , Antonio Maffei
Self-diagnosis functionalities, as integral components of advanced manufacturing services within cyber–physical systems (CPSs), are made possible through cloud computing technologies and machine learning techniques. These services play a crucial role in enhancing the autonomy of CPSs and introducing cost-efficient and scalable solutions. Despite the promising outlook, a gap exists in the literature regarding the lack of clear architectural frameworks and requirements for implementing self-diagnosis services in industrial settings. This paper addresses this gap by presenting a comprehensive requirement set and developing a high-level architecture tailored for self-diagnosis services. The proposed approach is validated through a detailed case study of a cloud-based self-diagnosis service, demonstrating alignment with the established architecture and requirements. The anticipated outcome of this research is to offer concrete implementation guidelines to support researchers, engineers, and practitioners in deploying CPS-based self-diagnosis services and improving production processes and system performance.
{"title":"Self-diagnosis service to support analysis of production performance, monitoring and optimisation activities","authors":"José Joaquín Peralta Abadía , Fabio Marco Monetti , Sylvia Nathaly Rea Minango , Angela Carrera-Rivera , Miriam Ugarte Querejeta , Mikel Cuesta Zabaljauregui , Felix Larrinaga Barrenechea , Miren Illarramendi Rezabal , Antonio Maffei","doi":"10.1016/j.jmsy.2025.11.010","DOIUrl":"10.1016/j.jmsy.2025.11.010","url":null,"abstract":"<div><div>Self-diagnosis functionalities, as integral components of advanced manufacturing services within cyber–physical systems (CPSs), are made possible through cloud computing technologies and machine learning techniques. These services play a crucial role in enhancing the autonomy of CPSs and introducing cost-efficient and scalable solutions. Despite the promising outlook, a gap exists in the literature regarding the lack of clear architectural frameworks and requirements for implementing self-diagnosis services in industrial settings. This paper addresses this gap by presenting a comprehensive requirement set and developing a high-level architecture tailored for self-diagnosis services. The proposed approach is validated through a detailed case study of a cloud-based self-diagnosis service, demonstrating alignment with the established architecture and requirements. The anticipated outcome of this research is to offer concrete implementation guidelines to support researchers, engineers, and practitioners in deploying CPS-based self-diagnosis services and improving production processes and system performance.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 800-821"},"PeriodicalIF":14.2,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516962","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-10DOI: 10.1016/j.jmsy.2025.11.003
Yanghao Wu, Paul D. Wilcox, Anthony J. Croxford
Fasteners are widely used in mechanical structures, where stress concentrations around fastener holes can lead to crack initiation and fatigue failures. In the aerospace industry, routine fastener hole inspections are critical to ensure structural integrity. Ultrasonic testing is one of the main inspection approaches. Conventionally, it involves a single-element probe that must be manually placed at multiple locations and orientations so that the ultrasound beam insonifies the area around the hole from different angles. The received ultrasonic time-domain signals at each location are analyzed, which is time-consuming, operator-dependent, and prone to inconsistencies. 2D ultrasonic array probes enable 3D volumetric images of fastener hole defects to be obtained from a single probe position, offering the potential for more efficient automated inspection and data interpretation. To achieve this, the 2D array probe must be accurately located over the centre of the hole and the ultrasonic coupling with the component must be consistent over the entire probe contact surface. This paper presents an automated robotic system for ultrasonic fastener hole inspection, that is designed to address these issues. A 7 degree-of-freedom robot arm is used with a vision module, and a customized probe adapter integrates a 2D ultrasonic array and coupling block into the robot end effector. A novel hybrid probe manipulation method is proposed, which combines camera-based visual localization with real-time ultrasound signal feedback to ensure accurate probe alignment and consistent coupling. The whole inspection workflow is scheduled and a graphical user interface is developed to demonstrate this automatic inspection. Experimental validation demonstrates that the robotic system performs accurate, repeatable inspections, significantly enhancing efficiency and reliability compared to manual techniques. The proposed approach addresses key challenges in robotic ultrasonic inspection and offers a scalable solution for intelligent maintenance in aerospace and other high-reliability industries.
{"title":"Robotic inspection of fastener holes with hybrid visual and ultrasonic motion control","authors":"Yanghao Wu, Paul D. Wilcox, Anthony J. Croxford","doi":"10.1016/j.jmsy.2025.11.003","DOIUrl":"10.1016/j.jmsy.2025.11.003","url":null,"abstract":"<div><div>Fasteners are widely used in mechanical structures, where stress concentrations around fastener holes can lead to crack initiation and fatigue failures. In the aerospace industry, routine fastener hole inspections are critical to ensure structural integrity. Ultrasonic testing is one of the main inspection approaches. Conventionally, it involves a single-element probe that must be manually placed at multiple locations and orientations so that the ultrasound beam insonifies the area around the hole from different angles. The received ultrasonic time-domain signals at each location are analyzed, which is time-consuming, operator-dependent, and prone to inconsistencies. 2D ultrasonic array probes enable 3D volumetric images of fastener hole defects to be obtained from a single probe position, offering the potential for more efficient automated inspection and data interpretation. To achieve this, the 2D array probe must be accurately located over the centre of the hole and the ultrasonic coupling with the component must be consistent over the entire probe contact surface. This paper presents an automated robotic system for ultrasonic fastener hole inspection, that is designed to address these issues. A 7 degree-of-freedom robot arm is used with a vision module, and a customized probe adapter integrates a 2D ultrasonic array and coupling block into the robot end effector. A novel hybrid probe manipulation method is proposed, which combines camera-based visual localization with real-time ultrasound signal feedback to ensure accurate probe alignment and consistent coupling. The whole inspection workflow is scheduled and a graphical user interface is developed to demonstrate this automatic inspection. Experimental validation demonstrates that the robotic system performs accurate, repeatable inspections, significantly enhancing efficiency and reliability compared to manual techniques. The proposed approach addresses key challenges in robotic ultrasonic inspection and offers a scalable solution for intelligent maintenance in aerospace and other high-reliability industries.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 770-783"},"PeriodicalIF":14.2,"publicationDate":"2025-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145516963","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-08DOI: 10.1016/j.jmsy.2025.11.004
Zhongji Su , Zexi Hua , Yongchuan Tang , Qingyuan Zhu , Zhipeng Qi , Lei Wang
Uncertainty in maintenance timing affects planning for systems with time window constraints, creating risks of overflow and operational disruptions. This paper proposes a multi-objective robust optimization method for Cluster system maintenance planning, integrating Glue Value at Risk (GlueVaR) to capture timing uncertainty. The method restores system reliability through maintenance while using GlueVaR to quantify timing uncertainty. Using GlueVaR's multi-parameter features to capture decision-makers' risk preferences, the method embeds maintenance strategies and decision tendencies into system metrics. The approach constructs a multi-objective optimization model with nested maintenance-level decisions and task scheduling. An improved Multi-Objective Evolutionary Algorithm based on Decomposition (MOEA/D) solves the model, screens optimal solutions, and analyzes time window overflow risk. Simulations on equipment clusters from outdoor signaling systems at railway stations show that maintenance risks decrease by 31.13 %, 45.54 %, and 61.09 % under generally optimistic, relatively conservative, and conservative decision-making tendencies, respectively. These results confirm the correctness and effectiveness of the proposed methodology.
维护时间的不确定性影响了有时间窗口限制的系统规划，造成了溢出和操作中断的风险。本文提出了一种多目标鲁棒优化方法，利用Glue Value at Risk （GlueVaR）来捕获时间不确定性。该方法通过维护恢复系统可靠性，同时使用GlueVaR量化时序不确定性。该方法利用GlueVaR的多参数特征捕捉决策者的风险偏好，将维护策略和决策倾向嵌入到系统度量中。该方法构建了一个具有嵌套维护级决策和任务调度的多目标优化模型。采用改进的基于分解的多目标进化算法（MOEA/D）对模型进行求解，筛选最优解，分析时间窗溢出风险。对火车站室外信号系统设备群的仿真结果表明，在一般乐观、相对保守和保守决策倾向下，维修风险分别降低了31.13 %、45.54 %和61.09 %。这些结果证实了所提出方法的正确性和有效性。
{"title":"Cluster system maintenance scheduling multi-objective optimization method integrating time uncertainty GlueVaR risk","authors":"Zhongji Su , Zexi Hua , Yongchuan Tang , Qingyuan Zhu , Zhipeng Qi , Lei Wang","doi":"10.1016/j.jmsy.2025.11.004","DOIUrl":"10.1016/j.jmsy.2025.11.004","url":null,"abstract":"<div><div>Uncertainty in maintenance timing affects planning for systems with time window constraints, creating risks of overflow and operational disruptions. This paper proposes a multi-objective robust optimization method for Cluster system maintenance planning, integrating Glue Value at Risk (GlueVaR) to capture timing uncertainty. The method restores system reliability through maintenance while using GlueVaR to quantify timing uncertainty. Using GlueVaR's multi-parameter features to capture decision-makers' risk preferences, the method embeds maintenance strategies and decision tendencies into system metrics. The approach constructs a multi-objective optimization model with nested maintenance-level decisions and task scheduling. An improved Multi-Objective Evolutionary Algorithm based on Decomposition (MOEA/D) solves the model, screens optimal solutions, and analyzes time window overflow risk. Simulations on equipment clusters from outdoor signaling systems at railway stations show that maintenance risks decrease by 31.13 %, 45.54 %, and 61.09 % under generally optimistic, relatively conservative, and conservative decision-making tendencies, respectively. These results confirm the correctness and effectiveness of the proposed methodology.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 752-769"},"PeriodicalIF":14.2,"publicationDate":"2025-11-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145465406","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Energy-efficient and data-driven decision-making has become a critical priority in modern manufacturing, particularly in customized or make-to-order (MTO) production where product variability causes large fluctuations in power consumption. Existing prediction models in this domain are often deterministic, lacking the ability to quantify uncertainty and capture hierarchical data dependencies, which limits their reliability for operational use. This study addresses this gap by developing a hierarchical Bayesian learning framework for power consumption prediction in customized stainless-steel manufacturing. The objective is to design models that not only achieve high predictive accuracy but also provide calibrated uncertainty estimates to support risk-aware production decisions. Four models, i.e., Hierarchical Bayesian Linear Regression (HBLR), Hierarchical Bayesian Neural Network (HBNN), Fully Connected Neural Network (FCN), and One-Dimensional Convolutional Neural Network (1D-CNN), were implemented and benchmarked using three inference algorithms: No-U-Turn Sampler (NUTS), Automatic Differentiation Variational Inference (ADVI), and Stein Variational Gradient Descent (SVGD). The innovation lies in systematically quantifying uncertainty using coverage probability, sharpness, and calibration error, and in establishing a unified comparison between probabilistic and deterministic models. Results show that the HBLR–NUTS model achieves the best trade-off between accuracy (RMSE = 11.85) and calibration quality (coverage 0.98), while ADVI offers near-equivalent performance with significantly lower computation time. These uncertainty-aware predictions can be directly integrated into Manufacturing Execution System (MES) and Enterprise Resource Planning (ERP) environments for energy-optimized scheduling and cost-aware planning. The proposed framework provides a scalable, interpretable, and statistically reliable foundation for advancing sustainable, data-driven manufacturing analytics.
{"title":"Uncertainty-aware power consumption prediction in customized stainless-steel manufacturing: A comparative study of hierarchical Bayesian and deep neural models","authors":"Akarawint Chawalitanont, Atit Bashyal, Hendro Wicaksono","doi":"10.1016/j.jmsy.2025.10.010","DOIUrl":"10.1016/j.jmsy.2025.10.010","url":null,"abstract":"<div><div>Energy-efficient and data-driven decision-making has become a critical priority in modern manufacturing, particularly in customized or make-to-order (MTO) production where product variability causes large fluctuations in power consumption. Existing prediction models in this domain are often deterministic, lacking the ability to quantify uncertainty and capture hierarchical data dependencies, which limits their reliability for operational use. This study addresses this gap by developing a hierarchical Bayesian learning framework for power consumption prediction in customized stainless-steel manufacturing. The objective is to design models that not only achieve high predictive accuracy but also provide calibrated uncertainty estimates to support risk-aware production decisions. Four models, i.e., Hierarchical Bayesian Linear Regression (HBLR), Hierarchical Bayesian Neural Network (HBNN), Fully Connected Neural Network (FCN), and One-Dimensional Convolutional Neural Network (1D-CNN), were implemented and benchmarked using three inference algorithms: No-U-Turn Sampler (NUTS), Automatic Differentiation Variational Inference (ADVI), and Stein Variational Gradient Descent (SVGD). The innovation lies in systematically quantifying uncertainty using coverage probability, sharpness, and calibration error, and in establishing a unified comparison between probabilistic and deterministic models. Results show that the HBLR–NUTS model achieves the best trade-off between accuracy (RMSE = 11.85) and calibration quality (coverage <span><math><mo>≈</mo></math></span> 0.98), while ADVI offers near-equivalent performance with significantly lower computation time. These uncertainty-aware predictions can be directly integrated into Manufacturing Execution System (MES) and Enterprise Resource Planning (ERP) environments for energy-optimized scheduling and cost-aware planning. The proposed framework provides a scalable, interpretable, and statistically reliable foundation for advancing sustainable, data-driven manufacturing analytics.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 713-735"},"PeriodicalIF":14.2,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145465401","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1016/j.jmsy.2025.11.002
Fuxuan Chi , Han Lin , Jinchuan Zheng , Baohua Jia
Roll-to-Roll (R2R) system is widely employed in continuous manufacturing processes requiring high throughput and precise control. Conventional tension-based control mechanism works well for most materials, but it is insufficient for high precision fabrication of optical fibers, which exhibit viscoelastic properties due to their polymer protective layer. Optical fibers can elongate without noticeable tension variation, significantly compromising the position accuracy, a serious issue in applications such as distributed optical sensing. To date, no existing control strategies in R2R systems have adequately addressed this limitation, nor have system models been developed that capture both fiber tension and elongation simultaneously. Here, we propose, to the best of our knowledge, a tension control scheme and a model for R2R systems to simultaneously account for fiber tension and elongation. The model incorporates the analysis of fiber viscoelastic deformation during winding, tension variation induced by winding, utilizing parameter identification and simplification approach developed through combined simulation and experiments. It is verified by experiments with two commonly used polymer-coated fibers, namely acrylic and polyimide, under diverse conditions, including different winding speeds and tension references. The experimental results confirm the model’s accuracy in predicting the elongation and tension variation of the fiber during the R2R winding process. By enabling system analysis and accurate prediction of material elongation, this model facilitates position-aimed pre-compensation in R2R systems, significantly enhancing position accuracy. It is applicable to a wide range of R2R processes for optical fibers (with or without viscoelasticity), such as in the fabrication of Fiber Bragg Grating (FBG) arrays.
{"title":"Modeling and optimization of positioning setpoints in a roll-to-roll system for optical fiber manufacturing","authors":"Fuxuan Chi , Han Lin , Jinchuan Zheng , Baohua Jia","doi":"10.1016/j.jmsy.2025.11.002","DOIUrl":"10.1016/j.jmsy.2025.11.002","url":null,"abstract":"<div><div>Roll-to-Roll (R2R) system is widely employed in continuous manufacturing processes requiring high throughput and precise control. Conventional tension-based control mechanism works well for most materials, but it is insufficient for high precision fabrication of optical fibers, which exhibit viscoelastic properties due to their polymer protective layer. Optical fibers can elongate without noticeable tension variation, significantly compromising the position accuracy, a serious issue in applications such as distributed optical sensing. To date, no existing control strategies in R2R systems have adequately addressed this limitation, nor have system models been developed that capture both fiber tension and elongation simultaneously. Here, we propose, to the best of our knowledge, a tension control scheme and a model for R2R systems to simultaneously account for fiber tension and elongation. The model incorporates the analysis of fiber viscoelastic deformation during winding, tension variation induced by winding, utilizing parameter identification and simplification approach developed through combined simulation and experiments. It is verified by experiments with two commonly used polymer-coated fibers, namely acrylic and polyimide, under diverse conditions, including different winding speeds and tension references. The experimental results confirm the model’s accuracy in predicting the elongation and tension variation of the fiber during the R2R winding process. By enabling system analysis and accurate prediction of material elongation, this model facilitates position-aimed pre-compensation in R2R systems, significantly enhancing position accuracy. It is applicable to a wide range of R2R processes for optical fibers (with or without viscoelasticity), such as in the fabrication of Fiber Bragg Grating (FBG) arrays.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 736-751"},"PeriodicalIF":14.2,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145465402","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-05DOI: 10.1016/j.jmsy.2025.10.015
Xinyu Cao , Binzi Xu , Dengchao Huang , Wei Li , Chun Wang , Maoshan Liu , Yan Wang
In current computer-aided process planning (CAPP) systems, the quality of the typical process routes employed directly influences the overall quality of subsequent process planning. With the advent of the big data era, automated analysis and discovery of typical process routes using advanced artificial intelligence (AI) techniques have become a critical issue to address. Current research primarily focuses on linear/simple process routes, with relatively limited exploration of networked process routes. Therefore, considering the characteristics of networked process routes, this paper proposes a novel approach for discovering typical networked process routes based on networked sequence similarity and intelligent clustering. Specifically, by thoroughly analyzing the information requirements of networked process routes and integrating five embedded process information types, a multi-dimensional process information fusion-based comprehensive similarity measure is constructed using the Kuhn–Munkres (KM) algorithm and principal component analysis (PCA). Furthermore, to ensure the clustering effectiveness of the discovered typical networked process routes, quantity and radius soft constraints are introduced into the traditional typical process route discovery problem. Two nutcracker optimization algorithm (NOA)-optimized affinity propagation (AP) algorithms (i.e., NOA-OAP and NOA-IAP) are proposed to address this problem, aiming to enhance clustering performance and identify more suitable and practical typical networked process routes for CAPP. Finally, numerical illustrations validate that the proposed similarity measure can effectively distinguish subtle differences among various networked process routes, and the two proposed clustering algorithms can discover more representative and effective typical process routes.
{"title":"A novel typical networked process route discovery approach based on networked sequence similarity and intelligent clustering","authors":"Xinyu Cao , Binzi Xu , Dengchao Huang , Wei Li , Chun Wang , Maoshan Liu , Yan Wang","doi":"10.1016/j.jmsy.2025.10.015","DOIUrl":"10.1016/j.jmsy.2025.10.015","url":null,"abstract":"<div><div>In current computer-aided process planning (CAPP) systems, the quality of the typical process routes employed directly influences the overall quality of subsequent process planning. With the advent of the big data era, automated analysis and discovery of typical process routes using advanced artificial intelligence (AI) techniques have become a critical issue to address. Current research primarily focuses on linear/simple process routes, with relatively limited exploration of networked process routes. Therefore, considering the characteristics of networked process routes, this paper proposes a novel approach for discovering typical networked process routes based on networked sequence similarity and intelligent clustering. Specifically, by thoroughly analyzing the information requirements of networked process routes and integrating five embedded process information types, a multi-dimensional process information fusion-based comprehensive similarity measure is constructed using the Kuhn–Munkres (KM) algorithm and principal component analysis (PCA). Furthermore, to ensure the clustering effectiveness of the discovered typical networked process routes, quantity and radius soft constraints are introduced into the traditional typical process route discovery problem. Two nutcracker optimization algorithm (NOA)-optimized affinity propagation (AP) algorithms (i.e., NOA-OAP and NOA-IAP) are proposed to address this problem, aiming to enhance clustering performance and identify more suitable and practical typical networked process routes for CAPP. Finally, numerical illustrations validate that the proposed similarity measure can effectively distinguish subtle differences among various networked process routes, and the two proposed clustering algorithms can discover more representative and effective typical process routes.</div></div>","PeriodicalId":16227,"journal":{"name":"Journal of Manufacturing Systems","volume":"83 ","pages":"Pages 687-712"},"PeriodicalIF":14.2,"publicationDate":"2025-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145465403","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号