Computers & Industrial Engineering最新文献

{"title":"Multi-dimensional fusion prediction and sensitivity analysis for overall equipment effectiveness based on temporal hybrid perception network","authors":"Di Zhou ,&nbsp;Jun Wang ,&nbsp;Ershun Pan ,&nbsp;Huimin Chen ,&nbsp;Zhaoxiang Chen","doi":"10.1016/j.cie.2025.111369","DOIUrl":"10.1016/j.cie.2025.111369","url":null,"abstract":"<div><div>Overall equipment effectiveness (OEE) is an important indicator for evaluating production line performance. It is composed of three dimensions: availability, performance efficiency, and quality. Previous prediction studies have mainly relied on statistical and classical machine learning (ML) methods. These studies focus primarily on overall indicators, with insufficient exploration of the three sub-dimensions. This may lead to a lack of targeted improvement measures and incomplete performance evaluations. To address these challenges, this paper proposes a novel temporal hybrid perception network (THPN) for predicting data from different dimensions. The OEE prediction value is obtained through multiplicative fusion. The time series decomposition (TSD) method is used to extract three modes from the data. Change points are detected by employing dynamic thresholds on the residual component, thereby enabling predictive maintenance of the production line. Based on the proposed network and detection method, a comprehensive multi-dimensional fusion prediction framework is established. It supports offline model training and online tasks. To further investigate the factors that most significantly impact OEE, an improved sensitivity analysis method is employed. Experiments are conducted on the hot rolling production line dataset. The results show that the proposed method achieves a prediction accuracy of 96.09%, representing a 3.74% improvement over the end-to-end prediction method. Moreover, the proposed model outperforms other compared models. Under the production conditions considered in this study, the availability dimension contributes the most to OEE.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"208 ","pages":"Article 111369"},"PeriodicalIF":6.7,"publicationDate":"2025-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144588294","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}

{"title":"Ramp-up planning for aircraft assembly production considering learning effect","authors":"Zhongkai Bao,&nbsp;Lu Chen","doi":"10.1016/j.cie.2025.111336","DOIUrl":"10.1016/j.cie.2025.111336","url":null,"abstract":"<div><div>China’s commercial aircraft manufacturing is entering into the ramp-up phase during which the cycle time is accelerated. This paper studies the production ramp-up planning problem for an aircraft final assembly line considering learning effect. The decision is to determine the cycle time acceleration strategy, with the objective to guarantee the production efficiency of the assembly line in the ramp-up phase. A mixed-integer linear programming model that incorporates a Gauss-DeJong learning curve is developed to formulate the problem. Variable bounds and valid cuts are tailored to enhance the model. An exact solution approach based on a logic-based Benders decomposition algorithm is proposed to solve the problem. Comparative experiments reveal a significant improvement in computational performance of the proposed approach compared to the commercial solver. Managerial insights are discussed by addressing the real aircraft final assembly line.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"208 ","pages":"Article 111336"},"PeriodicalIF":6.7,"publicationDate":"2025-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144572514","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}

{"title":"A machine learning based semi-automated framework for house of quality analysis","authors":"Madhumathi Ponnusamy ,&nbsp;Tushar Sonvani ,&nbsp;Ratri Parida ,&nbsp;Gaurav Nanda","doi":"10.1016/j.cie.2025.111371","DOIUrl":"10.1016/j.cie.2025.111371","url":null,"abstract":"<div><div>A good understanding of customer requirements and preferences is critical for designing effective and successful products and services. House of Quality (HOQ) is a structured approach used to identify key customer requirements, link the requirements to relevant product features and technical specifications, compare the features currently available in multiple brands of similar products, and determine the most important product design and quality aspects that the company needs to work on to meet customer needs. In this study, we proposed a semi-automated machine learning (ML) based framework for efficiently performing House of Quality analysis using large collections of customer reviews. We evaluated this framework by analyzing amazon customer reviews of two leading brands of refrigerator filters, A (5910 reviews) and B (6928 reviews). The ML-based framework used topic modelling to identify prominent customer requirements from the user reviews including customer satisfaction, price comparison, quality, product comparison, genuine parts, and installation instructions, which were used as the input for HOQ analysis. The topic model output was qualitatively analyzed along with technical product manuals to construct the HOQ matrix linking customer requirements with different product features. The results suggest that both firms have opportunities to better satisfy customer expectations, for example, Company A required improvements in durability, product comparability, and quality consistency. Furthermore, expert validation was conducted, and the results showed a good alignment between the expert judgements and LDA outputs thereby confirming the reliability and generalizability of the approach. Overall, the proposed ML-based approach for HOQ analysis can efficiently derive insights from large collections of customer reviews and ensure the company’s product design and development efforts are better aligned with consumer expectations, thus bolstering the company’s competitive positioning and revenues.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"208 ","pages":"Article 111371"},"PeriodicalIF":6.7,"publicationDate":"2025-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144581278","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}

{"title":"Open-End Bin Packing Problem with Conflicts","authors":"Ece Nur Balık,&nbsp;Ali Ekici,&nbsp;Milad Elyasi,&nbsp;Okan Örsan Özener","doi":"10.1016/j.cie.2025.111293","DOIUrl":"10.1016/j.cie.2025.111293","url":null,"abstract":"<div><div>In this paper, we study the <em>Open-End Bin Packing Problem with Conflicts</em> (OEBPPC), which is a combination of two variants of the well-known bin packing problem: <em>Open-End Bin Packing Problem</em> and <em>Bin Packing Problem with Conflicts</em>. In OEBPPC, the aim is to place a set of items into the least number of fixed-sized bins. The bin capacity is allowed to be exceeded only by the last item placed into the bin, and there exist conflicts between some item pairs. Conflicting items cannot be packed into the same bin. We develop a mathematical formulation and lower bounding procedures for the problem. As a solution approach, we propose a <em>General Variable Neighborhood Search Algorithm</em> (GVNS). We compare the performance of the proposed algorithm both against the lower bounds and other algorithms adapted to address OEBPPC from the literature. We observe that the proposed solution approach outperforms the best benchmark algorithm by a margin of 2.26% on average and provides solutions with an average gap of 8.28% (with respect to a lower bound).</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"208 ","pages":"Article 111293"},"PeriodicalIF":6.7,"publicationDate":"2025-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144570640","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}

{"title":"A new bi-level sustainable reverse logistics waste management problem with accelerated Benders decomposition method","authors":"Zhifei Wang ,&nbsp;Yankui Liu","doi":"10.1016/j.cie.2025.111341","DOIUrl":"10.1016/j.cie.2025.111341","url":null,"abstract":"<div><div>Waste management is a topic of global concern. Reverse logistics has received growing attention in recent literature, driven by various factors including the expansion of new environmental regulations, enhanced social responsibility, and economic benefits. To address this important issue, we attempt to build a reverse logistics waste management network. On the one hand, this study concentrates on the particular hierarchical relationships in the reverse logistics waste management network, addressing the integration of short-term (operational) decisions and long-term (strategic) decisions. On the other hand, uncertainty in waste disposal costs at disposal centers is characterized using a pair of uncertainty sets. We reformulate our bi-level robust reverse logistics waste management model as its equivalent mixed-integer programming model. Based on the structural characteristics of the resulting mixed-integer programming, a new tailored Benders decomposition algorithm is designed with acceleration strategies. Finally, a real-world case study in Mazandaran, Iran is addressed to analyze the performance and efficiency of our proposed model and Benders decomposition algorithm. The computational results indicate that our proposed waste management scheme exhibits superior performance and lower conservatism. (i) Compared to deterministic model, the scheme only requires an additional 12.81% cost to resist uncertainties. (ii) When compared to robust optimization model, the scheme reduces the robust price by 4.52%. (iii) The tailored Benders decomposition algorithm demonstrates higher solving efficiency, significantly reducing the running time by hundreds or even thousands of seconds as the problem size increases, and even overcoming the difficulties that cannot be solved within a specified time.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"208 ","pages":"Article 111341"},"PeriodicalIF":6.7,"publicationDate":"2025-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144581277","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}

{"title":"Fixed Topology Minimum-Length Trees with Neighborhoods: A Steiner tree based approach","authors":"Víctor Blanco ,&nbsp;Gabriel González ,&nbsp;Justo Puerto","doi":"10.1016/j.cie.2025.111331","DOIUrl":"10.1016/j.cie.2025.111331","url":null,"abstract":"<div><div>In this paper, we introduce the Fixed Topology Minimum-Length Tree with Neighborhood Problem, which aims to embed a rooted tree-shaped graph into a <span><math><mi>d</mi></math></span>-dimensional metric space while minimizing its total length provided that the nodes must be embedded to some restricted areas. This problem has significant applications in efficiently routing cables or pipelines in engineering designs. We propose novel mathematical optimization-based approaches to solve different versions of the problem based on the domain for the embedding. In cases where the embedding maps to a continuous space, we provide several Mixed Integer Nonlinear Optimization formulations. If the embedding is to a network, we derive a mixed integer linear programming formulation as well as a dimensionality reduction methodology that allows for solving larger problems in less CPU time. A data-driven methodology is also proposed to construct a proper network based on the instance of the problem. We report the results of a battery of computational experiments that validate our proposal.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111331"},"PeriodicalIF":6.7,"publicationDate":"2025-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144535453","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}

{"title":"Combinatorial weighted superposition attraction algorithm for solving multiple criteria decision-making problems","authors":"Adil Baykasoğlu,&nbsp;Özge Büyükdeveci","doi":"10.1016/j.cie.2025.111364","DOIUrl":"10.1016/j.cie.2025.111364","url":null,"abstract":"<div><div>In this study, a metaheuristic-based optimization approach is proposed in order to enhance applicability and effectiveness of a classical Multiple Criteria Decision Making method (MCDM) that is known as permutation method. The permutation method offers several advantages, such as robustness against the rank-reversal phenomenon and the ability to bypass normalization and aggregation of alternative scores across criteria, enabling it to effectively manage diverse data types. These advantages stem from its ability to directly compare alternatives within a given criteria set. However, the standard permutation method requires evaluating all possible permutations, which is computationally intensive and does not consider the magnitude of differences between alternatives in satisfying the criteria. The proposed approach introduces a new technique for calculating the ranking values of permutations by considering the magnitude of differences between alternatives. It also employs a parallel Weighted Superposition Attraction (WSA) algorithm to efficiently search for permutations, addressing these difficulties and identifying the optimal permutation of alternatives in MCDM problems. The proposed approach is evaluated on a range of real-world case studies, including agile methods assessment, laptop computer selection and fuzzy personnel selection, as well as on large-scale randomly generated problem instances. To demonstrate its effectiveness and validity, the method is also benchmarked against several well-established MCDM techniques and widely recognized metaheuristic algorithms.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"208 ","pages":"Article 111364"},"PeriodicalIF":6.7,"publicationDate":"2025-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144605780","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}

{"title":"Prediction of cardiovascular disease cases driven by air pollution data: A novel kernel-based Mixed-frequency data Sampling Nonlinear Grey Model","authors":"Yimeng An ,&nbsp;Yaoguo Dang ,&nbsp;Junjie Wang ,&nbsp;Yu Feng ,&nbsp;Shaowen Yang ,&nbsp;Son T. Mai","doi":"10.1016/j.cie.2025.111326","DOIUrl":"10.1016/j.cie.2025.111326","url":null,"abstract":"<div><div>Cardiovascular disease (CVD) is closely related to air pollution, so using air pollution data to drive CVD prediction has significant potential. However, the sampling frequency mismatch and nonlinear correlation between CVD statistical data and air pollution monitoring data make forecasting more challenging. This study aims to establish the nonlinear relationship between low-frequency sampling (lf-) CVD cases and high-frequency sampling (hf-) air quality index (AQI) via a novel proposed mixed-frequency sampling grey model, called the <em>kernel-based Mixed-frequency data Sampling Nonlinear Grey Model (k-MSNGM)</em>. In k-MSNGM, (1) monthly CVD case data and daily AQI data are directly inputted into both the linear and nonlinear parts without frequency synchronization, thereby minimizing potential information loss and enhancing prediction accuracy; (2) the polynomial weight functions are introduced in the linear part to address the frequency mismatch between lf-data and hf-data; and (3) the nonlinear part employs high-dimensional function mapping to handle mixed-frequency data based on the kernel trick theory, which is also effective for varying frequency mismatches. Empirical results from CVD case forecasting in Tianjin, China, demonstrate that k-MSNGM outperforms 13 state-of-the-art techniques in terms of prediction accuracy. We use AQI as a leading indicator for out-of-sample dynamic forecasting of future CVD cases, aiming to provide data-driven support for urban healthcare responses and medical resource allocations. This study is among the innovative attempts to incorporate the kernel trick into mixed-frequency modelling frameworks to capture nonlinear relationships for CVD case prediction.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"208 ","pages":"Article 111326"},"PeriodicalIF":6.7,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144572518","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}

{"title":"A deep reinforcement learning driven bi-population evolutionary optimization for precast concrete scheduling with production transportation","authors":"Yu Du ,&nbsp;Jun-qing Li ,&nbsp;Pei-yong Duan ,&nbsp;Xiao-xue Geng","doi":"10.1016/j.cie.2025.111354","DOIUrl":"10.1016/j.cie.2025.111354","url":null,"abstract":"<div><div>Precast concrete (PC) scheduling in prefabricated component production is essential for prefabricated building construction, where the transportation between different scaled factories and construction sites cannot be neglected. Additionally, multi-functional machines in factories can achieve flexible scheduling, enabling more efficient production for construction demand. Therefore, this study designs a bi-population based deep Q-network (B-DQN) with two cooperation populations to address the distributed flexible job shop scheduling problem with production transportation (DFJSPT) under PC manufacturing environment. Two objectives, i.e., makespan and total energy consumption, are minimized simultaneously. Firstly, in solution initialization, seven strategies concerning factory distribution, operation sequence, and machine assignment are built. Then, two deep Q-networks with 23 state features and 12 actions are designed to obtain better solutions, where DQN-G and DQN-L networks are to select global and local actions in global and local populations, respectively. In global and local actions, problem-specific and random heuristics are arranged to balance both exploitation and exploration of the B-DQN. Finally, dynamic switching mechanism enables cross-population solution migration to maintain evolutionary diversity. The comparison experiments with other competitive algorithms validates the effectiveness of the proposed approach in solving DFJSPT.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111354"},"PeriodicalIF":6.7,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144549709","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}

{"title":"Augmented reality-enabled knowledge management in industrial maintenance: the DILEAF framework","authors":"Wanting Mao ,&nbsp;Sara Scheffer ,&nbsp;Arnab Majumdar","doi":"10.1016/j.cie.2025.111363","DOIUrl":"10.1016/j.cie.2025.111363","url":null,"abstract":"<div><div>Effective maintenance in industrial operations relies on efficient management of task-critical knowledge, particularly in dynamic and unpredictable environments. Traditional knowledge management (KM) approaches face challenges in handling fragmented data, delivering procedural guidance, and adapting to evolving operational demands. To address these limitations, this study introduces the Data, Information, Learning, Engagement, Application, Feedback (DILEAF) framework, a structured KM model that integrates augmented reality (AR) as a digital enabler for improving knowledge capture, transfer, and application in industrial maintenance. By leveraging AR’s capabilities in real-time visualisation, interactive procedural guidance, and dynamic feedback, the DILEAF framework enhances user engagement and operational adaptability. The effectiveness of the framework is validated through a case study within a rolling stock organisation, where iterative experiments in both laboratory and field environments demonstrated improvements in task accuracy, real-time decision-making, and system adaptability. It was shown that AR overlays play a crucial role in enabling early error detection and correction, directly supporting the overall task success rate. Furthermore, 91 % of participants in the case study expressed satisfaction with the clarity and usefulness of the information presented via AR, underscoring the framework’s effectiveness in delivering task-relevant knowledge and supporting robust performance in maintenance scenarios. These findings illustrate that the DILEAF framework provides a system-informed and operationally structured approach to industrial KM, bridging theoretical KM principles with practical AR-driven implementations. This study establishes a scalable and adaptable foundation for integrating AR into industrial workflows, contributing to the advancement of digitalised maintenance strategies in industrial engineering.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"208 ","pages":"Article 111363"},"PeriodicalIF":6.7,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144572513","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}