Pub Date : 2025-12-18DOI: 10.1016/j.cie.2025.111772
Jiayi Liu , Xiaolong Zhang , Xiaofei Tu , Wenjun Xu , Zude Zhou
Robotic peg-hole disassembly is a common task in automated product disassembly, where compliant strategies are essential to prevent damage from excessive forces. Traditionally, recognizing the state of the peg and hole requires additional operations after disassembly. However, the forces and moments generated during the compliant disassembly process itself can be utilized for state recognition, eliminating the need for additional steps. This paper integrates compliant disassembly and state recognition of peg-hole into a single operation. A soft actor-critic algorithm is employed to minimize disassembly forces. Afterwards, a double Long Short-Term Memory Transformer algorithm is introduced to recognize the state of the peg and the hole. Experiments results show that the converged soft actor-critic model maintains disassembly forces at approximately 3 N, significantly outperforming other algorithms in force reduction. The double Long Short-Term Memory Transformer algorithm exhibits superior accuracy of 93.38 % in state recognition. This integrated approach improves end-of-life product recycling efficiency by combining operational synergy with enhanced performance.
{"title":"Integration of compliant disassembly strategy and state recognition for robotic peg-hole disassembly","authors":"Jiayi Liu , Xiaolong Zhang , Xiaofei Tu , Wenjun Xu , Zude Zhou","doi":"10.1016/j.cie.2025.111772","DOIUrl":"10.1016/j.cie.2025.111772","url":null,"abstract":"<div><div>Robotic peg-hole disassembly is a common task in automated product disassembly, where compliant strategies are essential to prevent damage from excessive forces. Traditionally, recognizing the state of the peg and hole requires additional operations after disassembly. However, the forces and moments generated during the compliant disassembly process itself can be utilized for state recognition, eliminating the need for additional steps. This paper integrates compliant disassembly and state recognition of peg-hole into a single operation. A soft actor-critic algorithm is employed to minimize disassembly forces. Afterwards, a double Long Short-Term Memory Transformer algorithm is introduced to recognize the state of the peg and the hole. Experiments results show that the converged soft actor-critic model maintains disassembly forces at approximately 3 N, significantly outperforming other algorithms in force reduction. The double Long Short-Term Memory Transformer algorithm exhibits superior accuracy of 93.38 % in state recognition. This integrated approach improves end-of-life product recycling efficiency by combining operational synergy with enhanced performance.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111772"},"PeriodicalIF":6.5,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145792185","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 : 2025-12-16DOI: 10.1016/j.cie.2025.111765
Sheng Jing , Wenhan Fu
Integrated circuit (IC) design is a crucial industry sitting at the upstream of the semiconductor supply chain by providing application-specific design services that enable manufacturers to make flexible and strategic decisions. For IC design service providers, research and development (R&D) capability is the core source of productivity, which are heavily dependent on the effectiveness of project management and the performance of manpower allocation.
In the current intensified market competition and rapid technological evolution, the efficient allocation of R&D resources, particularly manpower, has become essential for IC design industry to maintain sustainable growth. The ability to scientifically match multi-skilled teams with diverse project requirements, design complexities, and potential returns is a key factor in in strengthening the core competitiveness of IC design. However, existing studies have paid limited attention to data-driven approaches for portfolio planning and resource allocation..
To address this gap, this study proposes an integrated optimization approach for skill-aware and collaborative IC design portfolios decision that integrates revenue evaluation, project selection and team assignment optimization to improve the effectiveness and resilience of R&D resource. A multi-objective evolutionary algorithm with feasibility-prioritized search, intelligent repair mechanism and dynamic penalty strategy is developed to derive near-optimal portfolio decisions effectively.
To validate the proposed approach, an empirical study is conducted in an IC design service company. The results show that the proposed approach has good feasibility and capable of improving the project undertaking efficiency and overall profitability, while optimizing project management process. The study provides a data-driven and adaptive decision-support framework to enhance R&D efficiency and organizational resilience in IC design management.
{"title":"Integrated optimization approach for skill-aware and collaborative IC design portfolios decision to enhance R&D project effectiveness and resilience","authors":"Sheng Jing , Wenhan Fu","doi":"10.1016/j.cie.2025.111765","DOIUrl":"10.1016/j.cie.2025.111765","url":null,"abstract":"<div><div>Integrated circuit (IC) design is a crucial industry sitting at the upstream of the semiconductor supply chain by providing application-specific design services that enable manufacturers to make flexible and strategic decisions. For IC design service providers, research and development (R&D) capability is the core source of productivity, which are heavily dependent on the effectiveness of project management and the performance of manpower allocation.</div><div>In the current intensified market competition and rapid technological evolution, the efficient allocation of R&D resources, particularly manpower, has become essential for IC design industry to maintain sustainable growth. The ability to scientifically match multi-skilled teams with diverse project requirements, design complexities, and potential returns is a key factor in in strengthening the core competitiveness of IC design. However, existing studies have paid limited attention to data-driven approaches for portfolio planning and resource allocation..</div><div>To address this gap, this study proposes an integrated optimization approach for skill-aware and collaborative IC design portfolios decision that integrates revenue evaluation, project selection and team assignment optimization to improve the effectiveness and resilience of R&D resource. A multi-objective evolutionary algorithm with feasibility-prioritized search, intelligent repair mechanism and dynamic penalty strategy is developed to derive near-optimal portfolio decisions effectively.</div><div>To validate the proposed approach, an empirical study is conducted in an IC design service company. The results show that the proposed approach has good feasibility and capable of improving the project undertaking efficiency and overall profitability, while optimizing project management process. The study provides a data-driven and adaptive decision-support framework to enhance R&D efficiency and organizational resilience in IC design management.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111765"},"PeriodicalIF":6.5,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145792123","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 : 2025-12-16DOI: 10.1016/j.cie.2025.111761
Bailin Wang , Yihan Pei , Shuaipeng Yuan , Hongzhi Chen , Qing Liu , Zhuolun Zhang , Tieke Li
Cycle time in steelmaking-continuous casting is a critical indicator of production rhythm, directly impacting production planning and delivery commitment. However, accurate prediction is challenging due to the complex nature of industrial data, which are often multi-source, high-dimensional, noisy, and exhibit strong spatiotemporal correlations. Moreover, practical applications require predictive models that are both accurate and interpretable to support traceability and anomaly diagnosis. To address this, this study proposes a prediction framework that considers industrial data characteristics and model interpretability. First, categorical variables are numerically encoded, and noise is reduced through target encoding and data transformation. Next, features are synthesized to capture the spatiotemporal correlations of the production process, and key features are extracted using model-agnostic permutation importance. We then construct a deep neural network prediction model incorporating dropout and early stopping mechanisms. Finally, Shapley additive explanation is used to interpret the model, identify the key factors influencing cycle time, analyze the causes of abnormal prolongation, and provide management recommendations. The experimental results show that the proposed method outperforms not only classical ensemble learning and regularized regression models, but also recent cycle time prediction methods developed for manufacturing scenarios, achieving an average reduction of 15.4% in root-mean-square error and 13.5% in mean absolute error. In addition, simulation experiments demonstrate that the model captures implicit scheduling preferences embedded in real operations, enabling the predicted cycle times to support delivery-commitment planning and provide a quantitative reference for evaluating scheduling strategies.
{"title":"Steelmaking-continuous casting cycle time prediction method based on industrial data characteristics and model interpretability","authors":"Bailin Wang , Yihan Pei , Shuaipeng Yuan , Hongzhi Chen , Qing Liu , Zhuolun Zhang , Tieke Li","doi":"10.1016/j.cie.2025.111761","DOIUrl":"10.1016/j.cie.2025.111761","url":null,"abstract":"<div><div>Cycle time in steelmaking-continuous casting is a critical indicator of production rhythm, directly impacting production planning and delivery commitment. However, accurate prediction is challenging due to the complex nature of industrial data, which are often multi-source, high-dimensional, noisy, and exhibit strong spatiotemporal correlations. Moreover, practical applications require predictive models that are both accurate and interpretable to support traceability and anomaly diagnosis. To address this, this study proposes a prediction framework that considers industrial data characteristics and model interpretability. First, categorical variables are numerically encoded, and noise is reduced through target encoding and data transformation. Next, features are synthesized to capture the spatiotemporal correlations of the production process, and key features are extracted using model-agnostic permutation importance. We then construct a deep neural network prediction model incorporating dropout and early stopping mechanisms. Finally, Shapley additive explanation is used to interpret the model, identify the key factors influencing cycle time, analyze the causes of abnormal prolongation, and provide management recommendations. The experimental results show that the proposed method outperforms not only classical ensemble learning and regularized regression models, but also recent cycle time prediction methods developed for manufacturing scenarios, achieving an average reduction of 15.4% in root-mean-square error and 13.5% in mean absolute error. In addition, simulation experiments demonstrate that the model captures implicit scheduling preferences embedded in real operations, enabling the predicted cycle times to support delivery-commitment planning and provide a quantitative reference for evaluating scheduling strategies.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111761"},"PeriodicalIF":6.5,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145792184","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}
Making sure that city services and public spaces are designed to meet the needs of people with disabilities is a key part of building a modern, inclusive society. This study focuses on the growing need for transportation systems that are better suited to people with disabilities around the world. A novel approach is presented, categorizing wheelchair and attendant delivery, along with transport services, into four distinct types. The proposed mathematical model seamlessly integrates vehicle routing with two types of time windows, accommodating the intricate schedules of disabled individuals. Customers are empowered to request wheelchairs, attendants, or both, and can opt for transportation-only services. A pivotal feature of this model is the incorporation of shared vehicle usage, along with communal wheelchair sharing, optimizing resource utilization. The problem is formulated as a mixed integer non-linear mathematical optimization model, focusing on minimizing transportation costs. To tackle computational intricacies, the original non-linear model is transformed into an equivalent linear format. Addressing the problem’s NP-hardness, a heuristic approach based on column generation is employed. The model’s efficacy is robustly confirmed through experimental results, showcasing significant improvements in transportation efficiency and service provision. This approach opens a promising avenue for developing accessible transport systems, fostering a more inclusive society for disabled individuals globally.
{"title":"Optimizing urban transport for disabled individuals with shared wheelchair and attendant services: A column generation heuristic","authors":"Faraz Salehi, S.M.J. Mirzapour Al-e-Hashem, Kosar Abdollahi","doi":"10.1016/j.cie.2025.111740","DOIUrl":"10.1016/j.cie.2025.111740","url":null,"abstract":"<div><div>Making sure that city services and public spaces are designed to meet the needs of people with disabilities is a key part of building a modern, inclusive society. This study focuses on the growing need for transportation systems that are better suited to people with disabilities around the world. A novel approach is presented, categorizing wheelchair and attendant delivery, along with transport services, into four distinct types. The proposed mathematical model seamlessly integrates vehicle routing with two types of time windows, accommodating the intricate schedules of disabled individuals. Customers are empowered to request wheelchairs, attendants, or both, and can opt for transportation-only services. A pivotal feature of this model is the incorporation of shared vehicle usage, along with communal wheelchair sharing, optimizing resource utilization. The problem is formulated as a mixed integer non-linear mathematical optimization model, focusing on minimizing transportation costs. To tackle computational intricacies, the original non-linear model is transformed into an equivalent linear format. Addressing the problem’s NP-hardness, a heuristic approach based on column generation is employed. The model’s efficacy is robustly confirmed through experimental results, showcasing significant improvements in transportation efficiency and service provision. This approach opens a promising avenue for developing accessible transport systems, fostering a more inclusive society for disabled individuals globally.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111740"},"PeriodicalIF":6.5,"publicationDate":"2025-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145792121","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 : 2025-12-12DOI: 10.1016/j.cie.2025.111758
Tomislav Erdelić , Marko Đurasević , Martina Erdelić , Tonči Carić
Due to many regulations and policies aiming to reduce greenhouse gas emissions, logistic companies have started to incorporate electric vehicles within their vehicle fleet. To advance the electrification of delivery processes, in this paper, we investigate the Time-Dependent Electric Vehicle Routing Problem with Time Windows (TD-EVRPTW) and charging time dependent on the state of charge, considering time-dependent speeds and no delay time between service and departure times. We present the linearized Mixed Integer Linear Program (MILP) for the problem, and solve small instances by commercial software CPLEX. To solve large instances, we develop an Adaptive Large Neighborhood Search (ALNS) method. The ALNS method includes efficient approximation of changes in the solution for inter-route operators, exact procedure for charging station placement and local search for intensification. To validate the developed method, the solutions obtained by the ALNS approach are compared to state-of-the-art methods on related benchmark instances, achieving several best-known solutions. Results of ALNS on large instances indicate the benefits of time-dependent electric vehicle routing and provide managerial insights regarding the customer-specific attributes and effect of travel time linearization. Lastly, a real-world delivery problem is modeled as TD-EVRPTW, and solved by ALNS, showing the possible application of electric vehicles in time-dependent delivery processes.
{"title":"Time-Dependent Electric Vehicle Routing Problem with time windows and charging time dependent on the state of charge","authors":"Tomislav Erdelić , Marko Đurasević , Martina Erdelić , Tonči Carić","doi":"10.1016/j.cie.2025.111758","DOIUrl":"10.1016/j.cie.2025.111758","url":null,"abstract":"<div><div>Due to many regulations and policies aiming to reduce greenhouse gas emissions, logistic companies have started to incorporate electric vehicles within their vehicle fleet. To advance the electrification of delivery processes, in this paper, we investigate the Time-Dependent Electric Vehicle Routing Problem with Time Windows (TD-EVRPTW) and charging time dependent on the state of charge, considering time-dependent speeds and no delay time between service and departure times. We present the linearized Mixed Integer Linear Program (MILP) for the problem, and solve small instances by commercial software CPLEX. To solve large instances, we develop an Adaptive Large Neighborhood Search (ALNS) method. The ALNS method includes efficient <span><math><mrow><mi>O</mi><mrow><mo>(</mo><mn>1</mn><mo>)</mo></mrow></mrow></math></span> approximation of changes in the solution for inter-route operators, exact procedure for charging station placement and local search for intensification. To validate the developed method, the solutions obtained by the ALNS approach are compared to state-of-the-art methods on related benchmark instances, achieving several best-known solutions. Results of ALNS on large instances indicate the benefits of time-dependent electric vehicle routing and provide managerial insights regarding the customer-specific attributes and effect of travel time linearization. Lastly, a real-world delivery problem is modeled as TD-EVRPTW, and solved by ALNS, showing the possible application of electric vehicles in time-dependent delivery processes.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"213 ","pages":"Article 111758"},"PeriodicalIF":6.5,"publicationDate":"2025-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145766131","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 : 2025-12-11DOI: 10.1016/j.cie.2025.111755
Chao Li, Fan Yu, Qun Chen
This paper investigates a multi-depot collaborative vehicle routing problem with shared customers, delivery options, and flexible transshipment (MDCVRP-SCDOFT) in city logistics. In this problem, courier companies share delivery tasks and locker capacity, contributing to a more efficient logistics system while minimizing aggregate operating costs. We consider two customer types: those who prefer to collect their parcels from parcel lockers and those who require door-to-door delivery service. Parcel lockers serve dual purposes as both customer collection points and transshipment nodes (TNs). At TNs, couriers can deposit parcels for subsequent collection and delivery by couriers from different depots. We propose an adaptive large neighborhood search algorithm with embedded local search to solve this problem efficiently. Novel destroy and repair operators are developed by exploiting the problem structure, while existing operators from the literature are adapted accordingly. The proposed method is evaluated on benchmark instances derived from Solomon datasets and compared against genetic algorithm (GA), demonstrating superior performance. Furthermore, sensitivity analysis is conducted across instances of varying scales to derive valuable managerial insights.
{"title":"Multi-depot collaborative vehicle route problem with shared customer, delivery options and flexible transshipment","authors":"Chao Li, Fan Yu, Qun Chen","doi":"10.1016/j.cie.2025.111755","DOIUrl":"10.1016/j.cie.2025.111755","url":null,"abstract":"<div><div>This paper investigates a multi-depot collaborative vehicle routing problem with shared customers, delivery options, and flexible transshipment (MDCVRP-SCDOFT) in city logistics. In this problem, courier companies share delivery tasks and locker capacity, contributing to a more efficient logistics system while minimizing aggregate operating costs. We consider two customer types: those who prefer to collect their parcels from parcel lockers and those who require door-to-door delivery service. Parcel lockers serve dual purposes as both customer collection points and transshipment nodes (TNs). At TNs, couriers can deposit parcels for subsequent collection and delivery by couriers from different depots. We propose an adaptive large neighborhood search algorithm with embedded local search to solve this problem efficiently. Novel destroy and repair operators are developed by exploiting the problem structure, while existing operators from the literature are adapted accordingly. The proposed method is evaluated on benchmark instances derived from Solomon datasets and compared against genetic algorithm (GA), demonstrating superior performance. Furthermore, sensitivity analysis is conducted across instances of varying scales to derive valuable managerial insights.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"212 ","pages":"Article 111755"},"PeriodicalIF":6.5,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145748297","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 : 2025-12-10DOI: 10.1016/j.cie.2025.111752
Eyüp Ensar Işık , Şebnem Demirkol Akyol , Adil Baykasoğlu
Scheduling of sports competitions has become an area of interest for researchers with the globalization of sports and its spread to large crowds. Round Robin tournament derivatives are generally applied among different types of tournaments, especially in leagues. This study aims to create schedules for the Turkish Professional Football League. Indicators such as the number of breaks, weighted carry-over effect, and specific requirements are considered as characteristic values reflecting the league’s quality. In this study, an integer programming (IP) model is developed to solve the scheduling problem, and it is observed that the IP model gives the optimum schedule for small-sized problems only. As a remedy, a two-phase heuristic solution procedure is proposed. The heuristic procedure first finds a pattern set and then constitutes the schedule. Different pattern sets and schedules are presented in the experimental results. The results show that the proposed heuristic method obtains the best schedules concerning the current schedule and the proposed IP models for various problem characteristics.
{"title":"A systematic approach to the sports scheduling problem for the Turkish professional football league","authors":"Eyüp Ensar Işık , Şebnem Demirkol Akyol , Adil Baykasoğlu","doi":"10.1016/j.cie.2025.111752","DOIUrl":"10.1016/j.cie.2025.111752","url":null,"abstract":"<div><div>Scheduling of sports competitions has become an area of interest for researchers with the globalization of sports and its spread to large crowds. Round Robin tournament derivatives are generally applied among different types of tournaments, especially in leagues. This study aims to create schedules for the Turkish Professional Football League. Indicators such as the number of breaks, weighted carry-over effect, and specific requirements are considered as characteristic values reflecting the league’s quality. In this study, an integer programming (IP) model is developed to solve the scheduling problem, and it is observed that the IP model gives the optimum schedule for small-sized problems only. As a remedy, a two-phase heuristic solution procedure is proposed. The heuristic procedure first finds a pattern set and then constitutes the schedule. Different pattern sets and schedules are presented in the experimental results. The results show that the proposed heuristic method obtains the best schedules concerning the current schedule and the proposed IP models for various problem characteristics.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"212 ","pages":"Article 111752"},"PeriodicalIF":6.5,"publicationDate":"2025-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145747493","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 : 2025-12-10DOI: 10.1016/j.cie.2025.111750
Marva Ajab , Babar Zaman , Faraz Mukhtiar , Naveed Razzaq Butt , Muhammad Iftikhar Faraz
Statistical process control (SPC) is a critical tool in quality control that ensures uniform production standards. Control charts (CCs) are fundamental tools in SPC, used to track process performance and detect out-of-control behavior in production outputs. The cumulative sum (CUSUM) CC is particularly effective for detecting small to moderate shifts in process parameters. While the normality assumption is often adopted for CC design, many real-world quality characteristics deviate from normality and may be high-dimensional or skewed, challenging the applicability of classical methods. The current study introduces CC, a new type of CC that embeds the isolation forest technique (IsoForest) in the classical CUSUM framework for statistical process monitoring. Rather than assuming any underlying distribution, this method calculates anomaly scores from isolation trees to identify persistent shifts in process location parameters, making it especially effective for non-normally distributed data. CC thresholds are determined through Monte Carlo simulations and their efficiency is assessed using the median run length as a key performance metric. Its advantage is particularly notable for moderate and small shifts under both symmetric (multivariate normal) and heavy-tailed (multivariate ) distributions. The main contribution of this study is integrating IsoForest with CUSUM to create a robust, distribution-free CC that outperforms existing methods in detecting persistent process shifts. Finally, a real-life example is provided to demonstrate the practical applicability of the proposed CC.
{"title":"Enhancing statistical process control with machine learning: The Iso-CUSUM control chart for multivariate process","authors":"Marva Ajab , Babar Zaman , Faraz Mukhtiar , Naveed Razzaq Butt , Muhammad Iftikhar Faraz","doi":"10.1016/j.cie.2025.111750","DOIUrl":"10.1016/j.cie.2025.111750","url":null,"abstract":"<div><div>Statistical process control (SPC) is a critical tool in quality control that ensures uniform production standards. Control charts (CCs) are fundamental tools in SPC, used to track process performance and detect out-of-control behavior in production outputs. The cumulative sum (CUSUM) CC is particularly effective for detecting small to moderate shifts in process parameters. While the normality assumption is often adopted for CC design, many real-world quality characteristics deviate from normality and may be high-dimensional or skewed, challenging the applicability of classical methods. The current study introduces <span><math><mrow><mi>I</mi><mi>s</mi><mi>o</mi><mtext>-CUSUM</mtext></mrow></math></span> CC, a new type of CC that embeds the isolation forest technique (IsoForest) in the classical CUSUM framework for statistical process monitoring. Rather than assuming any underlying distribution, this method calculates anomaly scores from isolation trees to identify persistent shifts in process location parameters, making it especially effective for non-normally distributed data. CC thresholds are determined through Monte Carlo simulations and their efficiency is assessed using the median run length as a key performance metric. Its advantage is particularly notable for moderate and small shifts under both symmetric (multivariate normal) and heavy-tailed (multivariate <span><math><mi>t</mi></math></span>) distributions. The main contribution of this study is integrating IsoForest with CUSUM to create a robust, distribution-free CC that outperforms existing methods in detecting persistent process shifts. Finally, a real-life example is provided to demonstrate the practical applicability of the proposed <span><math><mrow><mi>I</mi><mi>s</mi><mi>o</mi><mtext>-CUSUM</mtext></mrow></math></span> CC.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"212 ","pages":"Article 111750"},"PeriodicalIF":6.5,"publicationDate":"2025-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145747505","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 : 2025-12-09DOI: 10.1016/j.cie.2025.111745
Yingxue Ren , Menghua Huang , Min Zhang , Zhen He
The continuous growing demand from stakeholders has driven more dairy companies to adopt Corporate Social Responsibility (CSR) to improve their performance. However, whether and how CSR is associated with quality performance (QP) of dairy companies is still unclear and lacks evidence. Building on the stakeholder theory and contingency theory, our study developed a conceptual framework with hypotheses that CSR is associated with dairy companies, with Total Quality Management (TQM) linking the two. A digital survey was design and conducted to collect data. 127 respondents from four dairy companies and 356 respondents from other 17 dairy companies were approached for pilot study and formal study respectively. The findings reveal that CSR is positively associated with both QP and TQM. Specifically, employee responsibility (EM) and customer responsibility (CU) are significantly positively associated with QP and TQM of dairy companies, indicating that effective engagement with both internal and external stakeholders is critical. TQM is linked to the association between CSR and QP, suggesting that the positive association between CSR and QP is strengthened when CSR is combined with TQM. An emerging trend toward shared responsibility in China’s dairy industry—combining CSR-driven stakeholder collaboration with TQM’s systemic quality control—demonstrates how actors across the dairy value chain can coordinate effectively, highlighting CSR–TQM integration as a promising direction for future research. Our findings enrich the CSR and quality-management literature with empirical evidence from the distinctive context of China’s dairy sector and offer managerial implications for firms and related stakeholders seeking to strengthen CSR and quality practices to ensure food safety.
{"title":"Corporate social responsibility practices and quality performance among Chinese dairy companies: The mediating role of total quality management","authors":"Yingxue Ren , Menghua Huang , Min Zhang , Zhen He","doi":"10.1016/j.cie.2025.111745","DOIUrl":"10.1016/j.cie.2025.111745","url":null,"abstract":"<div><div>The continuous growing demand from stakeholders has driven more dairy companies to adopt Corporate Social Responsibility (CSR) to improve their performance. However, whether and how CSR is associated with quality performance (QP) of dairy companies is still unclear and lacks evidence. Building on the stakeholder theory and contingency theory, our study developed a conceptual framework with hypotheses that CSR is associated with dairy companies, with Total Quality Management (TQM) linking the two. A digital survey was design and conducted to collect data. 127 respondents from four dairy companies and 356 respondents from other 17 dairy companies were approached for pilot study and formal study respectively. The findings reveal that CSR is positively associated with both QP and TQM. Specifically, employee responsibility (EM) and customer responsibility (CU) are significantly positively associated with QP and TQM of dairy companies, indicating that effective engagement with both internal and external stakeholders is critical. TQM is linked to the association between CSR and QP, suggesting that the positive association between CSR and QP is strengthened when CSR is combined with TQM. An emerging trend toward shared responsibility in China’s dairy industry—combining CSR-driven stakeholder collaboration with TQM’s systemic quality control—demonstrates how actors across the dairy value chain can coordinate effectively, highlighting CSR–TQM integration as a promising direction for future research. Our findings enrich the CSR and quality-management literature with empirical evidence from the distinctive context of China’s dairy sector and offer managerial implications for firms and related stakeholders seeking to strengthen CSR and quality practices to ensure food safety.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"212 ","pages":"Article 111745"},"PeriodicalIF":6.5,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145747492","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 : 2025-12-08DOI: 10.1016/j.cie.2025.111743
Zehua Fei , Yueyan Li
The existing manufacturing process of prefabricated buildings, which depends on long-distance material transport and energy-intensive processes, is unable to reduce the carbon emissions of the construction industry effectively. The distributed manufacturing process for prefabricated buildings, which is modeled as a carbon-efficient integrated distributed heterogeneous no-wait flow shop scheduling problem (CEDHFSP), is investigated to minimize makespan and total carbon emissions. The distributed production process is designed to alleviate the pressing contradiction between efficient production and reduced carbon emissions in prefabricated building manufacturing. An inverse reinforcement learning driven cooperative optimization framework (IRLCOF) is proposed in this paper to address CEDHFSP. The cooperative initialization method is designed to generate the initial population. The metaheuristic algorithm, inverse reinforcement learning, and Q-learning mechanism are introduced to explore solution space. The properties of CEDHFSP are summarized as knowledge employed in IRLCOF. The experiments are implemented to illustrate that the performance of IRLCOF outperforms the state-of-the-art algorithm. Specifically, IRLCOF is at least 20% better than other comparison algorithms for solving CEDHFSP in the Inverse Generational Distance metric.
{"title":"Inverse reinforcement learning driven cooperative optimization framework for carbon-efficiency integrated shop scheduling: A prefabricated building perspective","authors":"Zehua Fei , Yueyan Li","doi":"10.1016/j.cie.2025.111743","DOIUrl":"10.1016/j.cie.2025.111743","url":null,"abstract":"<div><div>The existing manufacturing process of prefabricated buildings, which depends on long-distance material transport and energy-intensive processes, is unable to reduce the carbon emissions of the construction industry effectively. The distributed manufacturing process for prefabricated buildings, which is modeled as a carbon-efficient integrated distributed heterogeneous no-wait flow shop scheduling problem (CEDHFSP), is investigated to minimize makespan and total carbon emissions. The distributed production process is designed to alleviate the pressing contradiction between efficient production and reduced carbon emissions in prefabricated building manufacturing. An inverse reinforcement learning driven cooperative optimization framework (IRLCOF) is proposed in this paper to address CEDHFSP. The cooperative initialization method is designed to generate the initial population. The metaheuristic algorithm, inverse reinforcement learning, and Q-learning mechanism are introduced to explore solution space. The properties of CEDHFSP are summarized as knowledge employed in IRLCOF. The experiments are implemented to illustrate that the performance of IRLCOF outperforms the state-of-the-art algorithm. Specifically, IRLCOF is at least 20% better than other comparison algorithms for solving CEDHFSP in the Inverse Generational Distance metric.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"212 ","pages":"Article 111743"},"PeriodicalIF":6.5,"publicationDate":"2025-12-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145747494","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