Pub Date : 2026-01-05DOI: 10.1016/j.compind.2025.104432
Tianming Ni , Wen Jiang , Huaguo Liang , Xiaoqing Wen , Mu Nie
Accurate detection of a wide range of defect patterns on wafers is crucial for enhancing chip yield and ensuring the reliability of semiconductor manufacturing systems. As this process becomes increasingly complex, new types of defects — referred to as unknown defects — emerge on wafers. Traditional pattern recognition methods struggle in this setting because limited samples are insufficient to effectively train deep learning models. Moreover, these models are prone to catastrophic forgetting when incrementally trained on new defect classes. To address these challenges, this paper proposes a method termed Few-Shot Class Contrastive Incremental Learning (FCCIL) for unknown wafer map defect detection. FCCIL integrates a contrastive learning network for distinguishing novel defect types and an incremental learning model for dynamic knowledge updating—both designed to mitigate catastrophic forgetting, thereby enabling the detection of unknown defects in wafer maps with limited data. Experimental results demonstrate a 4% improvement in forgetting resistance over state-of-the-art approaches, confirming the effectiveness of FCCIL in real-world semiconductor manufacturing scenarios.
{"title":"Incremental learning strategies for improved detection of unknown defects in wafer maps with limited samples","authors":"Tianming Ni , Wen Jiang , Huaguo Liang , Xiaoqing Wen , Mu Nie","doi":"10.1016/j.compind.2025.104432","DOIUrl":"10.1016/j.compind.2025.104432","url":null,"abstract":"<div><div>Accurate detection of a wide range of defect patterns on wafers is crucial for enhancing chip yield and ensuring the reliability of semiconductor manufacturing systems. As this process becomes increasingly complex, new types of defects — referred to as unknown defects — emerge on wafers. Traditional pattern recognition methods struggle in this setting because limited samples are insufficient to effectively train deep learning models. Moreover, these models are prone to catastrophic forgetting when incrementally trained on new defect classes. To address these challenges, this paper proposes a method termed Few-Shot Class Contrastive Incremental Learning (FCCIL) for unknown wafer map defect detection. FCCIL integrates a contrastive learning network for distinguishing novel defect types and an incremental learning model for dynamic knowledge updating—both designed to mitigate catastrophic forgetting, thereby enabling the detection of unknown defects in wafer maps with limited data. Experimental results demonstrate a 4% improvement in forgetting resistance over state-of-the-art approaches, confirming the effectiveness of FCCIL in real-world semiconductor manufacturing scenarios.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"175 ","pages":"Article 104432"},"PeriodicalIF":9.1,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145897225","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-26DOI: 10.1016/j.compind.2025.104431
Runda Jia , Fengyang Jiang , Ranmeng Lin , Jun Zheng , Dakuo He , Feng Yu
The thickening–dewatering process is an important stage in mineral industrial production, and improving its energy efficiency by optimizing energy consumption is a key research direction. However, there is a scarcity of studies on comprehensive optimization strategies for this process. To address this gap and reduce the energy efficiency index (EEI) in thickening–dewatering operations, this paper introduces reinforcement learning (RL) to the process. Since RL methods are prone to falling into local optima, we combine ensemble learning (EL) with RL. Based on the soft actor–critic (SAC) algorithm, which performs well in scheduling problems, we propose the ensemble SAC (ESAC) algorithm. In ESAC, each actor interacts with the environment using its own parameter set, and only the actions that yield the highest rewards are used to update the parameters of all actors. A weighted global loss function is also designed to prevent overestimation of the value network. Results show that the ESAC algorithm clearly outperforms benchmark RL algorithms, with EL effectively improving exploration efficiency and decision quality of RL. A multi-strategy ensemble helps to avoid local optima and optimize decision-making. Furthermore, when applied to the thickening–dewatering process of a gold hydrometallurgical plant, ESAC reduced the EEI by 44.77% compared to manual operation and increased the average underflow concentration by 9.57%.
{"title":"Ensemble reinforcement learning for optimizing the energy efficiency index in the thickening–dewatering process","authors":"Runda Jia , Fengyang Jiang , Ranmeng Lin , Jun Zheng , Dakuo He , Feng Yu","doi":"10.1016/j.compind.2025.104431","DOIUrl":"10.1016/j.compind.2025.104431","url":null,"abstract":"<div><div>The thickening–dewatering process is an important stage in mineral industrial production, and improving its energy efficiency by optimizing energy consumption is a key research direction. However, there is a scarcity of studies on comprehensive optimization strategies for this process. To address this gap and reduce the energy efficiency index (EEI) in thickening–dewatering operations, this paper introduces reinforcement learning (RL) to the process. Since RL methods are prone to falling into local optima, we combine ensemble learning (EL) with RL. Based on the soft actor–critic (SAC) algorithm, which performs well in scheduling problems, we propose the ensemble SAC (ESAC) algorithm. In ESAC, each actor interacts with the environment using its own parameter set, and only the actions that yield the highest rewards are used to update the parameters of all actors. A weighted global loss function is also designed to prevent overestimation of the value network. Results show that the ESAC algorithm clearly outperforms benchmark RL algorithms, with EL effectively improving exploration efficiency and decision quality of RL. A multi-strategy ensemble helps to avoid local optima and optimize decision-making. Furthermore, when applied to the thickening–dewatering process of a gold hydrometallurgical plant, ESAC reduced the EEI by 44.77% compared to manual operation and increased the average underflow concentration by 9.57%.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"175 ","pages":"Article 104431"},"PeriodicalIF":9.1,"publicationDate":"2025-12-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840753","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-23DOI: 10.1016/j.compind.2025.104429
Timothy O. Olawumi , Stephen Ojo , Saheed Toyin Muftaudeen , Acheme Okolobia Odeh , Taiwo Amoo
Blockchain technology (BCT) holds significant potential to transform construction supply chains (CSCs) by addressing longstanding challenges related to transparency, efficiency, and traceability. This study investigates and develops a rigorous, KPI-centric framework that systematically maps blockchain’s enabling capabilities (ECs) to key performance indicators (KPIs) critical to CSC performance. Through a hybrid methodology combining content analysis and design science research (DSR), the paper introduces a web-based Decision Support Tool (DST) to guide stakeholders in evaluating the technical suitability of blockchain for construction projects. The DST operates in two phases: first, assessing blockchain applicability through a structured diagnostic; second, recommending ‘best-fit’ blockchain stacks by aligning selected KPIs with relevant use cases and ECs. Validation via simulated case scenarios demonstrates the DST’s robustness in supporting early-stage, technically grounded decision-making and recommends blockchain solutions tailored to user-defined KPIs and use cases. The findings reveal that BCT, through automation, immutable data sharing, decentralized governance, and the like, can significantly improve CSCs' performance. By bridging the gap between conceptual promise and practical application, this research provides both theoretical advancements and actionable insights for digital transformation in the construction industry. It contributes a replicable decision-support architecture for technology adoption and performance optimization in complex, multi-stakeholder supply chain environments.
{"title":"Assessing blockchain technology's technical utility in construction supply chains: A multi-KPI decision support approach via use cases","authors":"Timothy O. Olawumi , Stephen Ojo , Saheed Toyin Muftaudeen , Acheme Okolobia Odeh , Taiwo Amoo","doi":"10.1016/j.compind.2025.104429","DOIUrl":"10.1016/j.compind.2025.104429","url":null,"abstract":"<div><div>Blockchain technology (BCT) holds significant potential to transform construction supply chains (CSCs) by addressing longstanding challenges related to transparency, efficiency, and traceability. This study investigates and develops a rigorous, KPI-centric framework that systematically maps blockchain’s enabling capabilities (ECs) to key performance indicators (KPIs) critical to CSC performance. Through a hybrid methodology combining content analysis and design science research (DSR), the paper introduces a web-based Decision Support Tool (DST) to guide stakeholders in evaluating the <em>technical suitability</em> of blockchain for construction projects. The DST operates in two phases: first, assessing blockchain applicability through a structured diagnostic; second, recommending ‘best-fit’ blockchain stacks by aligning selected KPIs with relevant use cases and ECs. Validation via simulated case scenarios demonstrates the DST’s robustness in supporting early-stage, technically grounded decision-making and recommends blockchain solutions tailored to user-defined KPIs and use cases. The findings reveal that BCT, through automation, immutable data sharing, decentralized governance, and the like, can significantly improve CSCs' performance. By bridging the gap between conceptual promise and practical application, this research provides both theoretical advancements and actionable insights for digital transformation in the construction industry. It contributes a replicable decision-support architecture for technology adoption and performance optimization in complex, multi-stakeholder supply chain environments.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"175 ","pages":"Article 104429"},"PeriodicalIF":9.1,"publicationDate":"2025-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145823142","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-18DOI: 10.1016/j.compind.2025.104428
Yi Gu , Sizhong Qin , Wenjie Liao , Xinzheng Lu
Designing the component dimensions of reinforced concrete (RC) frame structures is a crucial aspect of structural design. However, the reliance on manual expertise results in low design efficiency and unstable quality. The use of heuristic optimization and artificial intelligence algorithms such as generative adversarial networks (GANs) and graph neural networks (GNNs) can enhance design quality and efficiency. However, heuristic optimization algorithms are slow, and the accuracy of GANs and GNNs is insufficient. Therefore, this study proposes a diffusion model-based method called frame-dimension diffusion for predicting the component dimensions in RC frame structures. By integrating multichannel masking and gradient-weighted correction, this model enhances the precision and robustness of the component dimension predictions for beams, columns, and slabs. Furthermore, a new dataset construction method is introduced that captures the key standard story features and seismic conditions to facilitate the learning process of the diffusion model. Through comprehensive experimental evaluations and case studies, the effectiveness of the proposed method has been demonstrated. Compared to heterogeneous GNNs, the prediction accuracy has improved by 33 %. Additionally, the inter-story drift ratio results align with engineer-designed specifications, and the material usage error is within 1 %.
{"title":"Intelligent design of dimensions of reinforced concrete frame structure components using diffusion models","authors":"Yi Gu , Sizhong Qin , Wenjie Liao , Xinzheng Lu","doi":"10.1016/j.compind.2025.104428","DOIUrl":"10.1016/j.compind.2025.104428","url":null,"abstract":"<div><div>Designing the component dimensions of reinforced concrete (RC) frame structures is a crucial aspect of structural design. However, the reliance on manual expertise results in low design efficiency and unstable quality. The use of heuristic optimization and artificial intelligence algorithms such as generative adversarial networks (GANs) and graph neural networks (GNNs) can enhance design quality and efficiency. However, heuristic optimization algorithms are slow, and the accuracy of GANs and GNNs is insufficient. Therefore, this study proposes a diffusion model-based method called frame-dimension diffusion for predicting the component dimensions in RC frame structures. By integrating multichannel masking and gradient-weighted correction, this model enhances the precision and robustness of the component dimension predictions for beams, columns, and slabs. Furthermore, a new dataset construction method is introduced that captures the key standard story features and seismic conditions to facilitate the learning process of the diffusion model. Through comprehensive experimental evaluations and case studies, the effectiveness of the proposed method has been demonstrated. Compared to heterogeneous GNNs, the prediction accuracy has improved by 33 %. Additionally, the inter-story drift ratio results align with engineer-designed specifications, and the material usage error is within 1 %.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"175 ","pages":"Article 104428"},"PeriodicalIF":9.1,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145785013","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","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.compind.2025.104418
Yijun Geng , Jianzhou Wang , Jinze Li , Zhiwu Li
Wind energy has significant potential owing to the continuous growth of wind power and advancements in technology. However, the evolution of wind speed is influenced by the complex interaction of multiple factors, making it highly variable. The nonlinear and nonstationary nature of wind speed evolution can have a considerable impact on the overall power system. To address this challenge, we propose an integrated multiframe wind speed prediction system based on fuzzy feature extraction. This system employs a convex subset partitioning approach using a triangular affiliation function for fuzzy feature extraction. By applying soft clustering to the subsets, constructing an affiliation matrix, and identifying clustering centers, the system introduces the concepts of inner and boundary domains. It subsequently calculates the distances from data points to the clustering centers by measuring both interclass and intraclass distances. This method updates the cluster centers using the membership matrix, generating optimal feature values. Building on this foundation, we use multiple machine learning methods to input the fuzzy features into the prediction model and integrate learning techniques to predict feature values. Because different datasets require different modeling approaches, the integrated weight-updating module was used to dynamically adjust model weights by setting a dual objective function to ensure the accuracy and stability of the prediction. The effectiveness of the proposed model in terms of prediction performance and generalization ability is demonstrated through an empirical analysis of data from the Penglai wind farm in China.
{"title":"A short-term integrated wind speed prediction system based on fuzzy set feature extraction and intelligent optimization","authors":"Yijun Geng , Jianzhou Wang , Jinze Li , Zhiwu Li","doi":"10.1016/j.compind.2025.104418","DOIUrl":"10.1016/j.compind.2025.104418","url":null,"abstract":"<div><div>Wind energy has significant potential owing to the continuous growth of wind power and advancements in technology. However, the evolution of wind speed is influenced by the complex interaction of multiple factors, making it highly variable. The nonlinear and nonstationary nature of wind speed evolution can have a considerable impact on the overall power system. To address this challenge, we propose an integrated multiframe wind speed prediction system based on fuzzy feature extraction. This system employs a convex subset partitioning approach using a triangular affiliation function for fuzzy feature extraction. By applying soft clustering to the subsets, constructing an affiliation matrix, and identifying clustering centers, the system introduces the concepts of inner and boundary domains. It subsequently calculates the distances from data points to the clustering centers by measuring both interclass and intraclass distances. This method updates the cluster centers using the membership matrix, generating optimal feature values. Building on this foundation, we use multiple machine learning methods to input the fuzzy features into the prediction model and integrate learning techniques to predict feature values. Because different datasets require different modeling approaches, the integrated weight-updating module was used to dynamically adjust model weights by setting a dual objective function to ensure the accuracy and stability of the prediction. The effectiveness of the proposed model in terms of prediction performance and generalization ability is demonstrated through an empirical analysis of data from the Penglai wind farm in China.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"175 ","pages":"Article 104418"},"PeriodicalIF":9.1,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145731202","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-02DOI: 10.1016/j.compind.2025.104419
Yannis Bertrand, Jochen De Weerdt, Estefanía Serral
Business processes (BPs) that are enhanced with Internet of Things (IoT) technology, such as smart manufacturing processes, leverage IoT devices like sensors to monitor and capture contextual data from the physical environments where processes are executed. While the execution of BPs is typically recorded in information systems as event logs, IoT-enhanced BPs also produce IoT data that can offer valuable contextual insights. However, existing process mining techniques, which typically focus on the control-flow perspective, often miss key insights into the dynamic interplay of process activity sequences and IoT data—such as how certain IoT readings may trigger or affect specific process activities. To address this gap, we propose TROPICCAL, a new technique for multi-perspective trace clustering that integrates three key perspectives: the control-flow perspective, the trace attribute data perspective, and the time series (TS) sensor data perspective. The main novelty of TROPICCAL is the analysis of so-called context events as part of the TS data perspective. These events mark process-significant happenings detected in the TS sensor data. Furthermore, in order to unravel more insights from the output of our technique, we propose approaches for cluster explainability based on permutation feature importance. We demonstrate the efficacy of our approach and compare it with the most related and advanced approaches from the literature using a real-world manufacturing use case. Expert evaluation through in-depth interviews reveals that TROPICCAL offers better insights into the multi-perspective variants of the process.
{"title":"TROPICCAL: Multi-perspective trace clustering for IoT-enhanced processes","authors":"Yannis Bertrand, Jochen De Weerdt, Estefanía Serral","doi":"10.1016/j.compind.2025.104419","DOIUrl":"10.1016/j.compind.2025.104419","url":null,"abstract":"<div><div>Business processes (BPs) that are enhanced with Internet of Things (IoT) technology, such as smart manufacturing processes, leverage IoT devices like sensors to monitor and capture contextual data from the physical environments where processes are executed. While the execution of BPs is typically recorded in information systems as event logs, IoT-enhanced BPs also produce IoT data that can offer valuable contextual insights. However, existing process mining techniques, which typically focus on the control-flow perspective, often miss key insights into the dynamic interplay of process activity sequences and IoT data—such as how certain IoT readings may trigger or affect specific process activities. To address this gap, we propose TROPICCAL, a new technique for multi-perspective trace clustering that integrates three key perspectives: the control-flow perspective, the trace attribute data perspective, and the time series (TS) sensor data perspective. The main novelty of TROPICCAL is the analysis of so-called <em>context events</em> as part of the TS data perspective. These events mark process-significant happenings detected in the TS sensor data. Furthermore, in order to unravel more insights from the output of our technique, we propose approaches for cluster explainability based on permutation feature importance. We demonstrate the efficacy of our approach and compare it with the most related and advanced approaches from the literature using a real-world manufacturing use case. Expert evaluation through in-depth interviews reveals that TROPICCAL offers better insights into the multi-perspective variants of the process.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"175 ","pages":"Article 104419"},"PeriodicalIF":9.1,"publicationDate":"2025-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145651743","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-20DOI: 10.1016/j.compind.2025.104414
Inyoung Oh , Gilsang Jang , Jinho Song , Moongu Son , Daewoon Kim , Junsang Yun , Kwanghee Ko
Mixed Reality (MR) technology integrates digital content with the real world to enable a cohesive user experience. Accurate pose estimation is crucial for aligning virtual content with physical surroundings, ensuring the virtual elements appear naturally in the user’s environment. This paper proposes a learning-based approach for accurate pose estimation using a monocular RGB (Red-Green-Blue) image, eliminating the need for markers and depth sensors. The method leverages YOLO6D (You Only Look Once Six-Dimensional) and a RoI (Region of Interest)-based color augmentation technique combined with Principal Component Analysis to enhance the accuracy of 6-DoF (Degrees of Freedom) pose estimation, while mitigating the effects of background variations and lighting changes. The proposed pose estimation method is incorporated into an MR-based remote collaboration framework, ensuring consistent and robust information rendering onto target objects across various devices. This integration enhances the reliability and effectiveness of MR-based remote collaboration. Experimental results demonstrate the superior performance of the proposed method, establishing it as a strong foundation for future MR-based remote collaboration frameworks.
{"title":"A mixed reality-based remote collaboration framework using improved pose estimation","authors":"Inyoung Oh , Gilsang Jang , Jinho Song , Moongu Son , Daewoon Kim , Junsang Yun , Kwanghee Ko","doi":"10.1016/j.compind.2025.104414","DOIUrl":"10.1016/j.compind.2025.104414","url":null,"abstract":"<div><div>Mixed Reality (MR) technology integrates digital content with the real world to enable a cohesive user experience. Accurate pose estimation is crucial for aligning virtual content with physical surroundings, ensuring the virtual elements appear naturally in the user’s environment. This paper proposes a learning-based approach for accurate pose estimation using a monocular RGB (Red-Green-Blue) image, eliminating the need for markers and depth sensors. The method leverages YOLO6D (You Only Look Once Six-Dimensional) and a RoI (Region of Interest)-based color augmentation technique combined with Principal Component Analysis to enhance the accuracy of 6-DoF (Degrees of Freedom) pose estimation, while mitigating the effects of background variations and lighting changes. The proposed pose estimation method is incorporated into an MR-based remote collaboration framework, ensuring consistent and robust information rendering onto target objects across various devices. This integration enhances the reliability and effectiveness of MR-based remote collaboration. Experimental results demonstrate the superior performance of the proposed method, establishing it as a strong foundation for future MR-based remote collaboration frameworks.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"174 ","pages":"Article 104414"},"PeriodicalIF":9.1,"publicationDate":"2025-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145560131","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-18DOI: 10.1016/j.compind.2025.104398
Peipei Ding , Shi Qiang Liu , Raymond Chiong , Sandeep Dhakal , Dewang Chen , Debiao Li , Hoi-Lam Ma , Sai-Ho Chung
A digital twin (DT) is a real-time, highly accurate, virtual replica that reflects the states and behaviours of physical objects or systems. DTs can enable monitoring, simulation, prediction, optimisation as well as the structured integration of technologies, data flows and functional processes within smart industries. In recent years, the DT technology has emerged as a research hotspot, which has prompted us to conduct a review of its development and application in various industries. We have identified 30 leading journals that have significantly contributed to DT research, with the Computers in Industry (CII) journal ranking second among these 30 journals with more than 80 related publications. After briefly discussing the key concepts and major milestones around the development and rapid adoption of DTs in smart industries, we focus on reviewing and analysing the DT publications from the CII journal from 2018 to present by systematically categorising them into four primary application domains: manufacturing, construction, transportation, and technologies and paradigms. We also discuss potential research opportunities (e.g., life cycle management, cross-disciplinary integration, human-machine collaboration) and challenges from a theoretical perspective, and provide managerial insights (e.g., building open standards, enhancing data access compatibility, extending DTs’ operational functions, applications to more industries) from a practical perspective. This review will be helpful for academic researchers and industrial practitioners to gain a broad understanding of the versatility of DTs, thereby fostering interdisciplinary innovation.
{"title":"A review of digital twins in smart industries: Concepts, milestones, trends, applications, opportunities and challenges","authors":"Peipei Ding , Shi Qiang Liu , Raymond Chiong , Sandeep Dhakal , Dewang Chen , Debiao Li , Hoi-Lam Ma , Sai-Ho Chung","doi":"10.1016/j.compind.2025.104398","DOIUrl":"10.1016/j.compind.2025.104398","url":null,"abstract":"<div><div>A digital twin (DT) is a real-time, highly accurate, virtual replica that reflects the states and behaviours of physical objects or systems. DTs can enable monitoring, simulation, prediction, optimisation as well as the structured integration of technologies, data flows and functional processes within smart industries. In recent years, the DT technology has emerged as a research hotspot, which has prompted us to conduct a review of its development and application in various industries. We have identified 30 leading journals that have significantly contributed to DT research, with the <em>Computers in Industry</em> (CII) journal ranking second among these 30 journals with more than 80 related publications. After briefly discussing the key concepts and major milestones around the development and rapid adoption of DTs in smart industries, we focus on reviewing and analysing the DT publications from the CII journal from 2018 to present by systematically categorising them into four primary application domains: manufacturing, construction, transportation, and technologies and paradigms. We also discuss potential research opportunities (e.g., life cycle management, cross-disciplinary integration, human-machine collaboration) and challenges from a theoretical perspective, and provide managerial insights (e.g., building open standards, enhancing data access compatibility, extending DTs’ operational functions, applications to more industries) from a practical perspective. This review will be helpful for academic researchers and industrial practitioners to gain a broad understanding of the versatility of DTs, thereby fostering interdisciplinary innovation.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"174 ","pages":"Article 104398"},"PeriodicalIF":9.1,"publicationDate":"2025-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145560052","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-18DOI: 10.1016/j.compind.2025.104417
Keke Zha, Jiabin Yuan, Lili Fan, Yiyu Shen, Xu Liu
Rock segmentation, a crucial task in deep space exploration, demands high algorithmic accuracy. However, existing high-precision deep learning models often suffer from high computational complexity and energy consumption, which limit their deployment in resource-constrained space environments. To address these challenges, we present the spike temporal residual transformer network (ST-RTNet), the first spike-driven rock segmentation model directly trained with spiking neural network (SNN) architectures. ST-RTNet integrates convolutional layers and Transformer modules, introducing a novel attention mechanism that incorporates the temporal dimension of SNNs. By leveraging neuron voltage dynamics over time, ST-RTNet captures both temporal and spatial information, thereby enhancing the precision of segmentation. We evaluate ST-RTNet on three datasets and compare its performance with recent rock segmentation models. Experiments demonstrate that ST-RTNet achieves up to 90.13% energy reduction on INT-8 chips and 83.76% on Float-32 chips compared to artificial neural network models, while maintaining competitive segmentation accuracy. These findings demonstrate that ST-RTNet provides an efficient and accurate solution for rock segmentation in space exploration.
{"title":"ST-RTNet: An energy-efficient spike temporal residual transformer network for rock segmentation in deep space exploration","authors":"Keke Zha, Jiabin Yuan, Lili Fan, Yiyu Shen, Xu Liu","doi":"10.1016/j.compind.2025.104417","DOIUrl":"10.1016/j.compind.2025.104417","url":null,"abstract":"<div><div>Rock segmentation, a crucial task in deep space exploration, demands high algorithmic accuracy. However, existing high-precision deep learning models often suffer from high computational complexity and energy consumption, which limit their deployment in resource-constrained space environments. To address these challenges, we present the spike temporal residual transformer network (ST-RTNet), the first spike-driven rock segmentation model directly trained with spiking neural network (SNN) architectures. ST-RTNet integrates convolutional layers and Transformer modules, introducing a novel attention mechanism that incorporates the temporal dimension of SNNs. By leveraging neuron voltage dynamics over time, ST-RTNet captures both temporal and spatial information, thereby enhancing the precision of segmentation. We evaluate ST-RTNet on three datasets and compare its performance with recent rock segmentation models. Experiments demonstrate that ST-RTNet achieves up to 90.13% energy reduction on INT-8 chips and 83.76% on Float-32 chips compared to artificial neural network models, while maintaining competitive segmentation accuracy. These findings demonstrate that ST-RTNet provides an efficient and accurate solution for rock segmentation in space exploration.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"174 ","pages":"Article 104417"},"PeriodicalIF":9.1,"publicationDate":"2025-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145560051","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-18DOI: 10.1016/j.compind.2025.104415
Siyue Yang, Qi Sima, Liang Shen, Yukun Bao
Electricity market liberalisation has heightened competitive pressure among retailers, necessitating the accurate forecasting of retail consumer demand to support informed strategic decision-making in markets. Within this landscape, electricity retailers face two critical challenges: expanding, dynamic customer bases with limited historical consumption data from newly enrolled customers; and heterogeneous consumption patterns across diverse consumers, which necessitates tailored analytical approaches. However, conventional local forecasting methods, which require building individual models for each consumer, become operationally inefficient, and it is practically impossible to use these to meet such challenges. Hence, this study proposes a decomposition-based multi-sight convolutional neural network as a unified global method to generate predictions for multiple consumers. Given the inherent periodicity in electricity consumption profiles, this model incorporates three modules to handle both the commonality and diversity of periodic features among different consumers simultaneously: (1) a built-in decomposition module to recognise universal periodic patterns, enabling generalisation to new customers through shared temporal variations; (2) temporal transformation from one-dimensional input sequences to two-dimensional space according to daily periodicity, representing temporal dependencies along both intra- and inter-day dimensions; (3) a novel multi-sight convolutional neural network block comprising parallel convolution branches specialised for diverse subregions of the two-dimensional tensors, effectively detecting and modelling heterogeneous consumer-specific periodic sub-patterns across multiple series. Experiments using real-world datasets demonstrate that the proposed model achieves superior performance for global forecasting tasks in terms of both prediction accuracy and computational efficiency, compared with advanced methods. Ablation studies validate the effectiveness of the designed architecture.
{"title":"Global electricity demand forecasting for multi-consumer retailers using a decomposition-based multi-sight convolutional neural network","authors":"Siyue Yang, Qi Sima, Liang Shen, Yukun Bao","doi":"10.1016/j.compind.2025.104415","DOIUrl":"10.1016/j.compind.2025.104415","url":null,"abstract":"<div><div>Electricity market liberalisation has heightened competitive pressure among retailers, necessitating the accurate forecasting of retail consumer demand to support informed strategic decision-making in markets. Within this landscape, electricity retailers face two critical challenges: expanding, dynamic customer bases with limited historical consumption data from newly enrolled customers; and heterogeneous consumption patterns across diverse consumers, which necessitates tailored analytical approaches. However, conventional local forecasting methods, which require building individual models for each consumer, become operationally inefficient, and it is practically impossible to use these to meet such challenges. Hence, this study proposes a decomposition-based multi-sight convolutional neural network as a unified global method to generate predictions for multiple consumers. Given the inherent periodicity in electricity consumption profiles, this model incorporates three modules to handle both the commonality and diversity of periodic features among different consumers simultaneously: (1) a built-in decomposition module to recognise universal periodic patterns, enabling generalisation to new customers through shared temporal variations; (2) temporal transformation from one-dimensional input sequences to two-dimensional space according to daily periodicity, representing temporal dependencies along both intra- and inter-day dimensions; (3) a novel multi-sight convolutional neural network block comprising parallel convolution branches specialised for diverse subregions of the two-dimensional tensors, effectively detecting and modelling heterogeneous consumer-specific periodic sub-patterns across multiple series. Experiments using real-world datasets demonstrate that the proposed model achieves superior performance for global forecasting tasks in terms of both prediction accuracy and computational efficiency, compared with advanced methods. Ablation studies validate the effectiveness of the designed architecture.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"174 ","pages":"Article 104415"},"PeriodicalIF":9.1,"publicationDate":"2025-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145560134","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","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