Computers in Industry最新文献_第8页

An online milling deformation prediction method for thin-walled features with domain adversarial neural networks under small samples 基于领域对抗神经网络的小样本薄壁特征铣削变形在线预测方法

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-08-18 DOI: 10.1016/j.compind.2025.104349

Jingjing Li , Guanghui Zhou , Chao Zhang , Zhijie Wei , Fengyi Lu

Thin-walled machining features are extensively utilized in the aerospace industry, where the milling deformation caused by their weak rigidity has been the most common quality concern. Efficient control of milling deformation for thin-walled features is essential for enhancing quality. However, the high cost and time-consuming nature of data collection for aviation parts, leading to a limited availability of process data, which presents a significant challenge for predicting deformation in aerospace components. To address this issue, this study aims to develop a high-precision milling deformation prediction method by fully leveraging the small-sample data from machining experiments and simulation data. This paper first constructs a thin-walled features deformation prediction framework by integrating Domain Adversarial Neural Networks (DANN) with a digital twin process model. Secondly, the DANN method is adopted to achieve online prediction of milling deformation for thin-walled features. A small quantity of experimental deformation data serves as the target domain for training dataset, whereas milling simulation data produced by finite element software serves as the source domain. Milling deformation is accurately predicted using adversarial training based on the DANN structure for domain regression and domain classification. The best results show that the proposed method achieves better goodness of fit under limited sample conditions, with a 5 % increase in the Coefficient of Determination (R²) and a 15 % reduction in Mean Absolute Error (MAE) compared to five baseline methods. In the end, the DANN approach was integrated into the digital twin system for the milling process, and a prototype system was constructed to demonstrate the viability of the suggested approach.

薄壁加工特性在航空航天工业中得到了广泛的应用，由于薄壁加工的弱刚性引起的铣削变形一直是航空航天工业中最常见的质量问题。薄壁件铣削变形的有效控制是提高加工质量的关键。然而，航空部件数据收集的高成本和耗时特性导致过程数据的可用性有限，这对预测航空部件的变形提出了重大挑战。为了解决这一问题，本研究旨在充分利用加工实验和仿真数据的小样本数据，开发高精度铣削变形预测方法。本文首先将领域对抗神经网络（DANN）与数字孪生过程模型相结合，构建了薄壁特征变形预测框架。其次，采用DANN方法实现了薄壁特征铣削变形的在线预测；少量的实验变形数据作为训练数据集的目标域，而有限元软件生成的铣削仿真数据作为源域。利用基于DANN结构的对抗训练进行领域回归和领域分类，准确预测铣削变形。最佳结果表明，该方法在有限样本条件下获得了更好的拟合优度，与五种基线方法相比，决定系数（R²）提高了5 %，平均绝对误差（MAE）降低了15 %。最后，将DANN方法集成到铣削过程的数字孪生系统中，并构建了一个原型系统来验证所建议方法的可行性。

{"title":"An online milling deformation prediction method for thin-walled features with domain adversarial neural networks under small samples","authors":"Jingjing Li , Guanghui Zhou , Chao Zhang , Zhijie Wei , Fengyi Lu","doi":"10.1016/j.compind.2025.104349","DOIUrl":"10.1016/j.compind.2025.104349","url":null,"abstract":"<div><div>Thin-walled machining features are extensively utilized in the aerospace industry, where the milling deformation caused by their weak rigidity has been the most common quality concern. Efficient control of milling deformation for thin-walled features is essential for enhancing quality. However, the high cost and time-consuming nature of data collection for aviation parts, leading to a limited availability of process data, which presents a significant challenge for predicting deformation in aerospace components. To address this issue, this study aims to develop a high-precision milling deformation prediction method by fully leveraging the small-sample data from machining experiments and simulation data. This paper first constructs a thin-walled features deformation prediction framework by integrating Domain Adversarial Neural Networks (DANN) with a digital twin process model. Secondly, the DANN method is adopted to achieve online prediction of milling deformation for thin-walled features. A small quantity of experimental deformation data serves as the target domain for training dataset, whereas milling simulation data produced by finite element software serves as the source domain. Milling deformation is accurately predicted using adversarial training based on the DANN structure for domain regression and domain classification. The best results show that the proposed method achieves better goodness of fit under limited sample conditions, with a 5 % increase in the Coefficient of Determination (R²) and a 15 % reduction in Mean Absolute Error (MAE) compared to five baseline methods. In the end, the DANN approach was integrated into the digital twin system for the milling process, and a prototype system was constructed to demonstrate the viability of the suggested approach.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"172 ","pages":"Article 104349"},"PeriodicalIF":9.1,"publicationDate":"2025-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144865067","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}

引用次数: 0

Developing a Digital Twin of part cooling in an injection moulding process through a Dynamic Mode Decomposition-Kalman Filter approach 采用动态模态分解-卡尔曼滤波方法建立了注塑过程中零件冷却的数字孪生

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-08-16 DOI: 10.1016/j.compind.2025.104333

Mandana Kariminejad , David Tormey , Marion McAfee

A framework for creating a Digital Twin for spatiotemporal process monitoring is proposed based on Dynamic Mode Decomposition and the Kalman filter (DMD-KF). Many material processes require optimisation of complex spatiotemporal dynamics which are difficult to monitor with limited sensor measurements at accessible locations in the process. The DMD-KF approach facilitates the extraction of a spatiotemporal dynamic model with minimal computation time from numerical simulations, integrated with real-time sensor measurements of the accessible states to correct model errors. The method is demonstrated for real-time spatiotemporal monitoring of component cooling in the injection moulding process. Injection Moulding is a high-volume manufacturing process, which faces challenges in dimensional precision due to shrinkage and warpage defects which manifest post-production due to the gradual relaxation of internal residual stresses. To prevent internal stresses, the component should be sufficiently free of significant temperature differentials prior to ejection from the mould. However, the inaccessible nature of the mould tool limits sensor access for monitoring of the cooling phase. Dynamic Mode Decomposition (DMD) allows a best fit, linear state–space model of the temperature dynamics of 3D spatial nodes of the component to be extracted from a computationally-intensive finite element model of the process. Using only two thermocouple measurements, integrated with the DMD model via a Kalman Filter (KF), allows for construction of the 3D temperature map of the component inside the mould in real time. Simulation of different processing scenarios highlights that even with a DMD model developed under significantly different process conditions than used in implementation, the KF corrections still effectively estimate the temperature of critical states with with a root mean square error of 1.4 °C. The DMD-KF approach shows high potential for real-time spatio-temporal monitoring and quality prediction across diverse manufacturing processes.

提出了一种基于动态模态分解和卡尔曼滤波（ddm - kf）的时空过程监测数字孪生框架。许多材料过程需要优化复杂的时空动态，这在过程中难以用有限的传感器测量来监测。DMD-KF方法有助于以最小的计算时间从数值模拟中提取时空动态模型，并与可访问状态的实时传感器测量相结合，以纠正模型误差。该方法可用于注塑过程中零件冷却的实时时空监测。注射成型是一种大批量制造工艺，由于内部残余应力逐渐松弛，在生产后期会出现收缩和翘曲缺陷，因此在尺寸精度方面面临挑战。为了防止内应力，在从模具中弹出之前，组件应该充分避免显著的温差。然而，模具的不可接近性限制了传感器对冷却阶段的监测。动态模式分解（DMD）允许从计算密集型的过程有限元模型中提取组件三维空间节点的温度动力学的最佳拟合线性状态空间模型。仅使用两个热电偶测量，通过卡尔曼滤波器（KF）与DMD模型集成，就可以实时构建模具内组件的3D温度图。对不同加工场景的模拟表明，即使在与实际应用中明显不同的工艺条件下开发的DMD模型，KF修正仍然有效地估计临界状态的温度，均方根误差为1.4°C。DMD-KF方法在不同制造过程的实时时空监测和质量预测方面显示出很高的潜力。

{"title":"Developing a Digital Twin of part cooling in an injection moulding process through a Dynamic Mode Decomposition-Kalman Filter approach","authors":"Mandana Kariminejad , David Tormey , Marion McAfee","doi":"10.1016/j.compind.2025.104333","DOIUrl":"10.1016/j.compind.2025.104333","url":null,"abstract":"<div><div>A framework for creating a Digital Twin for spatiotemporal process monitoring is proposed based on Dynamic Mode Decomposition and the Kalman filter (DMD-KF). Many material processes require optimisation of complex spatiotemporal dynamics which are difficult to monitor with limited sensor measurements at accessible locations in the process. The DMD-KF approach facilitates the extraction of a spatiotemporal dynamic model with minimal computation time from numerical simulations, integrated with real-time sensor measurements of the accessible states to correct model errors. The method is demonstrated for real-time spatiotemporal monitoring of component cooling in the injection moulding process. Injection Moulding is a high-volume manufacturing process, which faces challenges in dimensional precision due to shrinkage and warpage defects which manifest post-production due to the gradual relaxation of internal residual stresses. To prevent internal stresses, the component should be sufficiently free of significant temperature differentials prior to ejection from the mould. However, the inaccessible nature of the mould tool limits sensor access for monitoring of the cooling phase. Dynamic Mode Decomposition (DMD) allows a best fit, linear state–space model of the temperature dynamics of 3D spatial nodes of the component to be extracted from a computationally-intensive finite element model of the process. Using only two thermocouple measurements, integrated with the DMD model via a Kalman Filter (KF), allows for construction of the 3D temperature map of the component inside the mould in real time. Simulation of different processing scenarios highlights that even with a DMD model developed under significantly different process conditions than used in implementation, the KF corrections still effectively estimate the temperature of critical states with with a root mean square error of 1.4 °C. The DMD-KF approach shows high potential for real-time spatio-temporal monitoring and quality prediction across diverse manufacturing processes.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"172 ","pages":"Article 104333"},"PeriodicalIF":9.1,"publicationDate":"2025-08-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144852480","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}

引用次数: 0

A platform to support the fast development of digital twins for agricultural holdings 支持农业控股数字孪生快速发展的平台

IF 1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-08-09 DOI: 10.1016/j.compind.2025.104347

Jorge Laguna, Mario E. Suaza-Medina, Rubén Béjar, Javier Lacasta, F. Javier Zarazaga-Soria

Industry 4.0 has advanced in agriculture through Smart Agriculture initiatives, yet open-field farming lags in the adoption of digital twins. Although digital twins have transformed manufacturing since 2011, their application in open-field farming remains limited by environmental variability, data scarcity, and financial constraints. This paper addresses four gaps: the lack of affordable platforms for small farms that dominate European agriculture; the need to manage agricultural complexity through data-driven models rather than the physical modelling approaches prevalent in non-agricultural sectors; the absence of open sources solutions adapted to agriculture’s slower innovation pace; the breach between technology and farmers. The platform features innovations in data workflow integration, open data incorporation, a cost-effective shared warehouse, and scalable data pipelines. To validate the proposed platform, a case study with two example digital twins mirroring two fields is conducted. This implementation ran efficiently on modest hardware (2 vCPUs, 4GB RAM). It achieved an average CPU usage of 60%, RAM usage of 2.5 GB, and a deployment time of around one minute. This helps lowering adoption barriers for small holdings and bridging the gap between basic monitoring and complex future systems.

通过智能农业计划，工业4.0在农业领域取得了进步，但露天农业在采用数字双胞胎方面落后。尽管自2011年以来，数字双胞胎已经改变了制造业，但它们在露天农业中的应用仍然受到环境变化、数据稀缺和财务约束的限制。本文解决了四个差距：缺乏主导欧洲农业的小农场负担得起的平台；需要通过数据驱动的模型来管理农业的复杂性，而不是非农业部门普遍采用的物理建模方法；缺乏适应农业创新步伐缓慢的开源解决方案；技术和农民之间的鸿沟。该平台在数据工作流集成、开放数据合并、具有成本效益的共享仓库和可扩展的数据管道方面进行了创新。为了验证所提出的平台，对两个镜像两个领域的示例数字双胞胎进行了案例研究。这个实现在适度的硬件（2个vcpu， 4GB RAM）上有效地运行。它的平均CPU使用率为60%，RAM使用率为2.5 GB，部署时间约为1分钟。这有助于降低小农场的采用障碍，并弥合基本监测与复杂的未来系统之间的差距。

{"title":"A platform to support the fast development of digital twins for agricultural holdings","authors":"Jorge Laguna, Mario E. Suaza-Medina, Rubén Béjar, Javier Lacasta, F. Javier Zarazaga-Soria","doi":"10.1016/j.compind.2025.104347","DOIUrl":"https://doi.org/10.1016/j.compind.2025.104347","url":null,"abstract":"Industry 4.0 has advanced in agriculture through Smart Agriculture initiatives, yet open-field farming lags in the adoption of digital twins. Although digital twins have transformed manufacturing since 2011, their application in open-field farming remains limited by environmental variability, data scarcity, and financial constraints. This paper addresses four gaps: the lack of affordable platforms for small farms that dominate European agriculture; the need to manage agricultural complexity through data-driven models rather than the physical modelling approaches prevalent in non-agricultural sectors; the absence of open sources solutions adapted to agriculture’s slower innovation pace; the breach between technology and farmers. The platform features innovations in data workflow integration, open data incorporation, a cost-effective shared warehouse, and scalable data pipelines. To validate the proposed platform, a case study with two example digital twins mirroring two fields is conducted. This implementation ran efficiently on modest hardware (2 vCPUs, 4GB RAM). It achieved an average CPU usage of 60%, RAM usage of 2.5 GB, and a deployment time of around one minute. This helps lowering adoption barriers for small holdings and bridging the gap between basic monitoring and complex future systems.","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"18 1","pages":""},"PeriodicalIF":10.0,"publicationDate":"2025-08-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144901449","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}

引用次数: 0

Flexibility-driven strategies for the optimal scheduling of industrial batteries across stacked energy markets 基于柔性驱动的能源市场工业电池优化调度策略

IF 9.1 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-07-30 DOI: 10.1016/j.compind.2025.104346

Adela Bâra, Simona-Vasilica Oprea

This paper presents a methodology that integrates price forecasting and optimization models to support decision-making for optimal trading strategies in the wholesale electricity market, specifically designed for Battery Energy Storage Systems (BESS). The objective is to maximize revenue from stacked market opportunities while minimizing capacity degradation caused by charge/discharge cycles and Depth of Discharge. The key contributions include embedding price forecasting models for Day-Ahead (DAM), Intra-Day (IDM) and Balancing Markets (BM), alongside a method to predict activation probability in the BM, allowing for better utilization of BESS capacity across multiple markets. Additionally, optimization models are proposed for trading across Ancillary Services Market (ASM), DAM, IDM and BM to enhance revenue. The models are integrated into a decision-support framework that evaluates various trading scenarios and selects the optimal strategy that maximizes future returns while minimizing degradation. Our research addresses the growing participation of investors in medium and large-scale BESS in electricity markets, where BESS play a role in capitalizing on price volatility, providing ancillary services and grid stability. The proposed methodology offers a comprehensive decision-support framework for investors to effectively optimize their trading decisions on multiple electricity markets. The study simulates a BESS with a 20 MWh capacity and 10 MW rated power on the Romanian wholesale electricity market from January to September 2024. Various trading scenarios are evaluated based on the number of charge/discharge cycles and reserved capacity cycles, with daily revenues assessed. The results show that trading scenarios with fewer energy cycles (typically one cycle) and multiple reserved capacity cycles (up to six) offer the best trade-off between revenue maximization and BESS degradation. Thus, the methodology extends the operational life of BESS, limiting degradation while maintaining profitability across energy markets.

本文提出了一种集成价格预测和优化模型的方法，以支持批发电力市场中最优交易策略的决策，专门为电池储能系统（BESS）设计。目标是从堆叠的市场机会中获得最大的收益，同时最大限度地减少由充放电周期和放电深度引起的容量下降。主要贡献包括嵌入日前（DAM）、日内（IDM）和平衡市场（BM）的价格预测模型，以及预测BM中激活概率的方法，从而更好地利用多个市场的BESS容量。此外，还提出了辅助服务市场（ASM）、DAM、IDM和BM之间交易的优化模型，以提高收益。这些模型被集成到一个决策支持框架中，该框架评估各种交易场景，并选择最优策略，使未来回报最大化，同时使退化最小化。我们的研究解决了投资者越来越多地参与电力市场的中型和大型BESS，其中BESS在利用价格波动，提供辅助服务和电网稳定方面发挥作用。该方法为投资者在多个电力市场上有效优化交易决策提供了一个全面的决策支持框架。该研究模拟了2024年1月至9月罗马尼亚批发电力市场上容量为20兆瓦时、额定功率为10 兆瓦的BESS。根据充放电周期和预留容量周期的数量评估各种交易场景，并评估每日收益。结果表明，较少的能源周期（通常为一个周期）和多个预留容量周期（最多六个）的交易场景在收益最大化和BESS退化之间提供了最佳折衷。因此，该方法延长了BESS的使用寿命，限制了退化，同时保持了能源市场的盈利能力。

{"title":"Flexibility-driven strategies for the optimal scheduling of industrial batteries across stacked energy markets","authors":"Adela Bâra, Simona-Vasilica Oprea","doi":"10.1016/j.compind.2025.104346","DOIUrl":"10.1016/j.compind.2025.104346","url":null,"abstract":"<div><div>This paper presents a methodology that integrates price forecasting and optimization models to support decision-making for optimal trading strategies in the wholesale electricity market, specifically designed for Battery Energy Storage Systems (BESS). The objective is to maximize revenue from stacked market opportunities while minimizing capacity degradation caused by charge/discharge cycles and Depth of Discharge. The key contributions include embedding price forecasting models for Day-Ahead (DAM), Intra-Day (IDM) and Balancing Markets (BM), alongside a method to predict activation probability in the BM, allowing for better utilization of BESS capacity across multiple markets. Additionally, optimization models are proposed for trading across Ancillary Services Market (ASM), DAM, IDM and BM to enhance revenue. The models are integrated into a decision-support framework that evaluates various trading scenarios and selects the optimal strategy that maximizes future returns while minimizing degradation. Our research addresses the growing participation of investors in medium and large-scale BESS in electricity markets, where BESS play a role in capitalizing on price volatility, providing ancillary services and grid stability. The proposed methodology offers a comprehensive decision-support framework for investors to effectively optimize their trading decisions on multiple electricity markets. The study simulates a BESS with a 20 MWh capacity and 10 MW rated power on the Romanian wholesale electricity market from January to September 2024. Various trading scenarios are evaluated based on the number of charge/discharge cycles and reserved capacity cycles, with daily revenues assessed. The results show that trading scenarios with fewer energy cycles (typically one cycle) and multiple reserved capacity cycles (up to six) offer the best trade-off between revenue maximization and BESS degradation. Thus, the methodology extends the operational life of BESS, limiting degradation while maintaining profitability across energy markets.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"172 ","pages":"Article 104346"},"PeriodicalIF":9.1,"publicationDate":"2025-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144738299","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}

引用次数: 0

A sensor-integrated digital twin framework for molten pool monitoring of laser powder bed fusion 激光粉末床熔融熔池监测的传感器集成数字孪生框架

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-07-16 DOI: 10.1016/j.compind.2025.104332

Xiaojun Peng , Renwu Jiang , Zhenghui Yuan , Aoming Zhang , Zhangdong Chen , Di Wang , Xiaoqi Chen , Yingjie Zhang

Digital twin technology has emerged as a transformative approach to enhance intelligent manufacturing processes. In the context of Laser Powder Bed Fusion (LPBF), this study presents a novel Sensor and Simulation Combined Digital Twin (SSC-DT) framework. The SSC-DT integrates high-speed camera sensor data and a reduced order model based on proper orthogonal decomposition with radial basis functions (POD-RBF) for the monitoring of critical molten pool characteristics, including width, depth, and mean temperature. The POD-RBF model enables rapid molten pool characteristics predictions. High-speed camera data is used to optimize the model parameters for accurate predictions. Validation against optical microscopy measurement shows that the SSC-DT achieves a mean relative error of 5.48% for molten pool width prediction and 5.68% for depth prediction, outperforming conventional models such as the Eagar-Tsai and Fabbro models. This framework demonstrates stable performance under conduction and slight keyhole modes, offering significant potential for intelligent control, predictive maintenance, and quality assurance in LPBF.

数字孪生技术已经成为增强智能制造过程的一种变革性方法。在激光粉末床融合（LPBF）的背景下，本研究提出了一种新的传感器和仿真相结合的数字孪生（SSC-DT）框架。SSC-DT将高速摄像传感器数据和基于适当正交分解的降阶模型与径向基函数（POD-RBF）相结合，用于监测熔池的关键特征，包括宽度、深度和平均温度。POD-RBF模型能够快速预测熔池特征。利用高速摄像机数据优化模型参数，实现准确预测。光学显微镜测量验证表明，SSC-DT预测熔池宽度的平均相对误差为5.48%，深度预测的平均相对误差为5.68%，优于Eagar-Tsai和Fabbro等传统模型。该框架在传导和轻微锁孔模式下表现出稳定的性能，为LPBF的智能控制、预测性维护和质量保证提供了巨大潜力。

{"title":"A sensor-integrated digital twin framework for molten pool monitoring of laser powder bed fusion","authors":"Xiaojun Peng , Renwu Jiang , Zhenghui Yuan , Aoming Zhang , Zhangdong Chen , Di Wang , Xiaoqi Chen , Yingjie Zhang","doi":"10.1016/j.compind.2025.104332","DOIUrl":"10.1016/j.compind.2025.104332","url":null,"abstract":"<div><div>Digital twin technology has emerged as a transformative approach to enhance intelligent manufacturing processes. In the context of Laser Powder Bed Fusion (LPBF), this study presents a novel Sensor and Simulation Combined Digital Twin (SSC-DT) framework. The SSC-DT integrates high-speed camera sensor data and a reduced order model based on proper orthogonal decomposition with radial basis functions (POD-RBF) for the monitoring of critical molten pool characteristics, including width, depth, and mean temperature. The POD-RBF model enables rapid molten pool characteristics predictions. High-speed camera data is used to optimize the model parameters for accurate predictions. Validation against optical microscopy measurement shows that the SSC-DT achieves a mean relative error of 5.48% for molten pool width prediction and 5.68% for depth prediction, outperforming conventional models such as the Eagar-Tsai and Fabbro models. This framework demonstrates stable performance under conduction and slight keyhole modes, offering significant potential for intelligent control, predictive maintenance, and quality assurance in LPBF.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"171 ","pages":"Article 104332"},"PeriodicalIF":8.2,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144657088","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}

引用次数: 0

Enhanced vision-based structural displacement monitoring through deep learning approaches 通过深度学习方法增强基于视觉的结构位移监测

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-07-10 DOI: 10.1016/j.compind.2025.104337

Yu Shanshan , Ziyang Su , Shuai Dong , Xiaoyuan He , Yaqiang Yang , Jian Zhang

Vision-based displacement monitoring for large-scale civil infrastructures remains challenged by limited imaging resolution and uncontrolled camera-motion. This study presents a hybrid deep learning (DL) framework addressing these dual challenges through two technical innovations. Firstly, we develop an enhanced video super-resolution (VSR) architecture based on BasicVSR++, incorporating a novel multi-scale feature extraction module with pre-alignment mechanism which uses a multi-stage bidirectional propagation strategy to optimize temporal feature fusion. Secondly, we devise a dual-stage convolutional neural networks (CNN) architecture for unsupervised homography (H) estimation, enabling coarse-to-fine camera motion compensation through parametric transformation. The integrated displacement measurement method combines super-resolved imagery with KAZE-DIC algorithm for sub-pixel target tracking under challenging conditions including low illumination, texture-deficient backgrounds, and camera-motion. Field validation on an 888-meter suspension bridge demonstrates the framework's potential for structural health monitoring applications. The proposed methodology advances vision-based metrology by simultaneously resolving resolution constraints and motion artifacts through synergistic DL strategies.

基于视觉的大型民用基础设施位移监测仍然面临成像分辨率有限和摄像机运动不受控制的挑战。本研究提出了一个混合深度学习（DL）框架，通过两项技术创新来解决这些双重挑战。首先，基于BasicVSR++开发了一种增强的视频超分辨率（VSR）架构，该架构结合了一种新型的多尺度特征提取模块和预对准机制，采用多阶段双向传播策略优化时间特征融合；其次，我们设计了一种双级卷积神经网络（CNN）架构用于无监督单应性(H)估计，通过参数变换实现从粗到精的相机运动补偿。集成位移测量方法将超分辨率图像与KAZE-DIC算法相结合，在低照度、纹理缺乏背景和相机运动等挑战性条件下实现亚像素目标跟踪。在一座888米的悬索桥上进行的现场验证证明了该框架在结构健康监测应用中的潜力。提出的方法通过协同深度学习策略同时解决分辨率约束和运动伪影，从而推进基于视觉的计量。

{"title":"Enhanced vision-based structural displacement monitoring through deep learning approaches","authors":"Yu Shanshan , Ziyang Su , Shuai Dong , Xiaoyuan He , Yaqiang Yang , Jian Zhang","doi":"10.1016/j.compind.2025.104337","DOIUrl":"10.1016/j.compind.2025.104337","url":null,"abstract":"<div><div>Vision-based displacement monitoring for large-scale civil infrastructures remains challenged by limited imaging resolution and uncontrolled camera-motion. This study presents a hybrid deep learning (DL) framework addressing these dual challenges through two technical innovations. Firstly, we develop an enhanced video super-resolution (VSR) architecture based on BasicVSR++, incorporating a novel multi-scale feature extraction module with pre-alignment mechanism which uses a multi-stage bidirectional propagation strategy to optimize temporal feature fusion. Secondly, we devise a dual-stage convolutional neural networks (CNN) architecture for unsupervised homography (H) estimation, enabling coarse-to-fine camera motion compensation through parametric transformation. The integrated displacement measurement method combines super-resolved imagery with KAZE-DIC algorithm for sub-pixel target tracking under challenging conditions including low illumination, texture-deficient backgrounds, and camera-motion. Field validation on an 888-meter suspension bridge demonstrates the framework's potential for structural health monitoring applications. The proposed methodology advances vision-based metrology by simultaneously resolving resolution constraints and motion artifacts through synergistic DL strategies.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"171 ","pages":"Article 104337"},"PeriodicalIF":8.2,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144595578","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}

引用次数: 0

From Ontologies to Knowledge Augmented Large Language Models for Automation: A decision-making guidance for achieving human–robot collaboration in Industry 5.0 从本体到面向自动化的知识增强大型语言模型：工业5.0中实现人机协作的决策指导

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-07-10 DOI: 10.1016/j.compind.2025.104329

John Oyekan , Christopher Turner , Michael Bax , Erich Graf

The rapid advancement of Large Language Models (LLMs) has resulted in interest in their potential applications within manufacturing systems, particularly in the context of Industry 5.0. However, determining when to implement LLMs versus other Natural Language Processing (NLP) techniques, ontologies or knowledge graphs, remains an open question. This paper offers decision-making guidance for selecting the most suitable technique in various industrial contexts, emphasizing human–robot collaboration and resilience in manufacturing. We examine the origins and unique strengths of LLMs, ontologies, and knowledge graphs, assessing their effectiveness across different industrial scenarios based on the number of domains or disciplines required to bring a product from design to manufacture. Through this comparative framework, we explore specific use cases where LLMs could enhance robotics for human–robot collaboration, while underscoring the continued relevance of ontologies and knowledge graphs in low-dependency or resource-constrained sectors. Additionally, we address the practical challenges of deploying these technologies, such as computational cost and interpretability, providing a roadmap for manufacturers to navigate the evolving landscape of Language based AI tools in Industry 5.0. Our findings offer a foundation for informed decision-making, helping industry professionals optimize the use of Language Based models for sustainable, resilient, and human-centric manufacturing. We also propose a Large Knowledge Language Model architecture that offers the potential for transparency and configuration based on complexity of task and computing resources available.

大型语言模型（llm）的快速发展引起了人们对其在制造系统中的潜在应用的兴趣，特别是在工业5.0的背景下。然而，确定何时实现llm与其他自然语言处理（NLP）技术、本体或知识图相比，仍然是一个悬而未决的问题。本文为在各种工业环境中选择最合适的技术提供了决策指导，强调了制造中的人机协作和弹性。我们研究了法学硕士、本体和知识图谱的起源和独特优势，根据产品从设计到制造所需的领域或学科数量，评估了它们在不同工业场景中的有效性。通过这个比较框架，我们探索了法学硕士可以增强人机协作的机器人技术的具体用例，同时强调了本体和知识图在低依赖性或资源受限领域的持续相关性。此外，我们还解决了部署这些技术的实际挑战，例如计算成本和可解释性，为制造商在工业5.0中导航基于语言的人工智能工具的不断发展的前景提供了路线图。我们的研究结果为明智的决策提供了基础，帮助行业专业人士优化基于语言的模型的使用，以实现可持续、有弹性和以人为中心的制造。我们还提出了一个大型知识语言模型体系结构，该体系结构提供了基于任务复杂性和可用计算资源的透明度和配置的潜力。

{"title":"From Ontologies to Knowledge Augmented Large Language Models for Automation: A decision-making guidance for achieving human–robot collaboration in Industry 5.0","authors":"John Oyekan , Christopher Turner , Michael Bax , Erich Graf","doi":"10.1016/j.compind.2025.104329","DOIUrl":"10.1016/j.compind.2025.104329","url":null,"abstract":"<div><div>The rapid advancement of Large Language Models (LLMs) has resulted in interest in their potential applications within manufacturing systems, particularly in the context of Industry 5.0. However, determining when to implement LLMs versus other Natural Language Processing (NLP) techniques, ontologies or knowledge graphs, remains an open question. This paper offers decision-making guidance for selecting the most suitable technique in various industrial contexts, emphasizing human–robot collaboration and resilience in manufacturing. We examine the origins and unique strengths of LLMs, ontologies, and knowledge graphs, assessing their effectiveness across different industrial scenarios based on the number of domains or disciplines required to bring a product from design to manufacture. Through this comparative framework, we explore specific use cases where LLMs could enhance robotics for human–robot collaboration, while underscoring the continued relevance of ontologies and knowledge graphs in low-dependency or resource-constrained sectors. Additionally, we address the practical challenges of deploying these technologies, such as computational cost and interpretability, providing a roadmap for manufacturers to navigate the evolving landscape of Language based AI tools in Industry 5.0. Our findings offer a foundation for informed decision-making, helping industry professionals optimize the use of Language Based models for sustainable, resilient, and human-centric manufacturing. We also propose a Large Knowledge Language Model architecture that offers the potential for transparency and configuration based on complexity of task and computing resources available.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"171 ","pages":"Article 104329"},"PeriodicalIF":8.2,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144588165","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}

引用次数: 0

SAMAC-R3-MED: Semantic alignment and multi-agent collaboration of retriever-reranker-responder models for multimodal engineering documents SAMAC-R3-MED：多模态工程文档检索者-重新排序者-应答者模型的语义对齐和多智能体协作

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-07-09 DOI: 10.1016/j.compind.2025.104336

Fei Li, Xinyu Li, Sijie Wen, Haoyang Huang, Jinsong Bao

In the manufacturing industry's lifecycle, a vast amount of engineering documents in text, table, and image formats is generated. Retrieval-augmented generation (RAG) models can enhance retrieval efficiency and adapt to evolving document knowledge. However, challenges in understanding multimodal semantic associations and the absence of engineering-semantic-aligned RAG models result in suboptimal accuracy. This paper introduces a novel approach, namely SAMAC-R³-MED, to tackle these challenges. First, a fine-grained context enhancement strategy is applied to multimodal large language models (MLLMs), bridging multimodal semantic understanding by constructing multi-modal semantic trees (MMST) and multi-modal knowledge graphs (MMKG), forming a hybrid retrieval base. Second, to bridge the semantic gap in RAG models, a new training framework, retriever-reranker-responder (R³), is proposed, utilizing supervised and reinforcement learning with ranking feedback to enhance alignment. Third, a multi-channel hybrid retrieval strategy is implemented for the multi-agent collaboration R³ models, integrating expert feedback, semantic trees, and graphs to optimize the RAG pipeline and improve the accuracy of retrieving multimodal associative semantic contexts. An engineering documents chat (eDoChat) system is implemented, in the case of wind turbine assembly, validating the effectiveness in retrieving and generating accurate multimodal answers. Ablation experiments show R³ models outperform traditional RAG models, and SAMAC-R³-MED achieves state-of-the-art results in multimodal retrieval and generation tasks.

在制造业的生命周期中，会生成大量文本、表格和图像格式的工程文档。检索增强生成（RAG）模型可以提高检索效率，适应不断变化的文档知识。然而，在理解多模态语义关联方面的挑战和缺乏与工程语义一致的RAG模型导致精度不理想。本文介绍了一种新的方法，即SAMAC-R3-MED，来解决这些挑战。首先，将细粒度上下文增强策略应用于多模态大语言模型（mllm），通过构建多模态语义树（MMST）和多模态知识图（MMKG）架起多模态语义理解的桥梁，形成混合检索库；其次，为了弥合RAG模型中的语义差距，提出了一种新的训练框架，即检索者-重新排序者-应答者（R3），利用有监督和强化学习与排名反馈来增强一致性。第三，针对多智能体协作R3模型，采用多通道混合检索策略，集成专家反馈、语义树和图，优化RAG管道，提高检索多模态关联语义上下文的准确性。以风力涡轮机装配为例，实现了一个工程文档聊天（eDoChat）系统，验证了检索和生成准确多模态答案的有效性。消融实验表明，R3模型优于传统的RAG模型，SAMAC-R3-MED在多模态检索和生成任务中取得了最先进的结果。

{"title":"SAMAC-R3-MED: Semantic alignment and multi-agent collaboration of retriever-reranker-responder models for multimodal engineering documents","authors":"Fei Li, Xinyu Li, Sijie Wen, Haoyang Huang, Jinsong Bao","doi":"10.1016/j.compind.2025.104336","DOIUrl":"10.1016/j.compind.2025.104336","url":null,"abstract":"<div><div>In the manufacturing industry's lifecycle, a vast amount of engineering documents in text, table, and image formats is generated. Retrieval-augmented generation (RAG) models can enhance retrieval efficiency and adapt to evolving document knowledge. However, challenges in understanding multimodal semantic associations and the absence of engineering-semantic-aligned RAG models result in suboptimal accuracy. This paper introduces a novel approach, namely SAMAC-R<sup>3</sup>-MED, to tackle these challenges. First, a fine-grained context enhancement strategy is applied to multimodal large language models (MLLMs), bridging multimodal semantic understanding by constructing multi-modal semantic trees (MMST) and multi-modal knowledge graphs (MMKG), forming a hybrid retrieval base. Second, to bridge the semantic gap in RAG models, a new training framework, retriever-reranker-responder (R<sup>3</sup>), is proposed, utilizing supervised and reinforcement learning with ranking feedback to enhance alignment. Third, a multi-channel hybrid retrieval strategy is implemented for the multi-agent collaboration R<sup>3</sup> models, integrating expert feedback, semantic trees, and graphs to optimize the RAG pipeline and improve the accuracy of retrieving multimodal associative semantic contexts. An engineering documents chat (eDoChat) system is implemented, in the case of wind turbine assembly, validating the effectiveness in retrieving and generating accurate multimodal answers. Ablation experiments show R<sup>3</sup> models outperform traditional RAG models, and SAMAC-R<sup>3</sup>-MED achieves state-of-the-art results in multimodal retrieval and generation tasks.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"171 ","pages":"Article 104336"},"PeriodicalIF":8.2,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144588054","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}

引用次数: 0

Moving-feature-driven label propagation for training data generation from target domains 从目标域生成训练数据的移动特征驱动标签传播

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-07-09 DOI: 10.1016/j.compind.2025.104335

Taegeon Kim , Wei-Chih Chern , Seokhwan Kim , Vijayan K. Asari , Hongjo Kim

Deep learning models often suffer from performance degradation when applied to construction sites that differ from the source domain due to their sensitivity to data distribution shifts. Although methods such as transfer learning, domain adaptation, and synthetic data generation have been explored to improve generalization, collecting and annotating data from new target domains remains a labor-intensive bottleneck. This study presents a self-training-based framework to generate training data for construction object detection in unlabeled target domains. The method identifies moving objects using optical flow estimation, propagates class labels through iterative self-training, and synthesizes realistic training images via image inpainting and copy-paste augmentation. Experimental results from four visually distinct construction scenes demonstrate that the proposed method significantly improves detection performance without relying on manually labeled target data. These findings contribute to advancing automated and scalable domain adaptation techniques for vision-based construction monitoring.

由于深度学习模型对数据分布变化的敏感性，当应用于与源域不同的建筑工地时，其性能往往会下降。虽然已经探索了迁移学习、领域适应和合成数据生成等方法来提高泛化，但从新的目标领域收集和注释数据仍然是一个劳动密集型的瓶颈。本文提出了一种基于自训练的框架，用于在未标记的目标域中生成用于建筑目标检测的训练数据。该方法利用光流估计识别运动目标，通过迭代自训练传播类标签，并通过图像绘制和复制粘贴增强合成逼真的训练图像。实验结果表明，该方法在不依赖人工标记目标数据的情况下显著提高了检测性能。这些发现有助于推进自动化和可扩展的领域适应技术，用于基于视觉的施工监测。

{"title":"Moving-feature-driven label propagation for training data generation from target domains","authors":"Taegeon Kim , Wei-Chih Chern , Seokhwan Kim , Vijayan K. Asari , Hongjo Kim","doi":"10.1016/j.compind.2025.104335","DOIUrl":"10.1016/j.compind.2025.104335","url":null,"abstract":"<div><div>Deep learning models often suffer from performance degradation when applied to construction sites that differ from the source domain due to their sensitivity to data distribution shifts. Although methods such as transfer learning, domain adaptation, and synthetic data generation have been explored to improve generalization, collecting and annotating data from new target domains remains a labor-intensive bottleneck. This study presents a self-training-based framework to generate training data for construction object detection in unlabeled target domains. The method identifies moving objects using optical flow estimation, propagates class labels through iterative self-training, and synthesizes realistic training images via image inpainting and copy-paste augmentation. Experimental results from four visually distinct construction scenes demonstrate that the proposed method significantly improves detection performance without relying on manually labeled target data. These findings contribute to advancing automated and scalable domain adaptation techniques for vision-based construction monitoring.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"171 ","pages":"Article 104335"},"PeriodicalIF":8.2,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144580088","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}

引用次数: 0

Attentive neural processes based on reliable inferences for industrial equipment anomaly detection 基于可靠推理的细心神经过程在工业设备异常检测中的应用

IF 8.2 1区计算机科学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computers in Industry

Pub Date : 2025-07-05 DOI: 10.1016/j.compind.2025.104331

Yuhang Huang , Bo Yang , Shilong Wang , Keqiang Xie , Yu Wang , Lili Yi , Nan Dong

In modern manufacturing, assessing equipment conditions has become increasingly costly due to the complexity of industrial machinery. While data-driven methods have partially addressed this challenge, traditional approaches are often limited by their assumption of a simple Gaussian data distribution. This assumption fails in high-dimensional, complex industrial scenarios, where traditional models cannot capture the true data distribution, reducing their effectiveness. This paper introduces a reliable inference attentive neural process (RIANP) based on normalizing flows (NFs) and neural ordinary differential equations (NODEs), a method for detecting anomalies in industrial equipment. NFs replace the fixed prior assumption of the attentive neural process (ANP) with a learnable prior distribution, addressing sampling holes caused by unlearnable priors. Next, NODEs model the posterior distribution, enabling smoother alignment between the learnable prior and complex posterior distributions. A validation of two industrial anomaly detection cases shows that the RIANP achieves an average F1 score of 94.64 %, a 7.5 % improvement over the ANP, and an AUC of 96.5 %, representing a 12 % enhancement.

在现代制造业中，由于工业机械的复杂性，评估设备状况的成本越来越高。虽然数据驱动的方法已经部分解决了这一挑战，但传统方法通常受到简单高斯数据分布假设的限制。这种假设在高维、复杂的工业场景中是行不通的，因为传统模型无法捕捉真实的数据分布，从而降低了它们的有效性。本文介绍了一种基于归一化流（NFs）和神经常微分方程（NODEs）的可靠推理注意神经过程（RIANP），用于工业设备的异常检测。NFs用可学习的先验分布取代了注意神经过程（ANP）的固定先验假设，解决了由不可学习先验引起的采样漏洞。接下来，节点对后验分布进行建模，使可学习的先验分布和复杂的后验分布之间的对齐更加平滑。对两个工业异常检测案例的验证表明，RIANP的平均F1得分为94.64 %，比ANP提高了7.5 %，AUC为96.5 %，提高了12 %。

{"title":"Attentive neural processes based on reliable inferences for industrial equipment anomaly detection","authors":"Yuhang Huang , Bo Yang , Shilong Wang , Keqiang Xie , Yu Wang , Lili Yi , Nan Dong","doi":"10.1016/j.compind.2025.104331","DOIUrl":"10.1016/j.compind.2025.104331","url":null,"abstract":"<div><div>In modern manufacturing, assessing equipment conditions has become increasingly costly due to the complexity of industrial machinery. While data-driven methods have partially addressed this challenge, traditional approaches are often limited by their assumption of a simple Gaussian data distribution. This assumption fails in high-dimensional, complex industrial scenarios, where traditional models cannot capture the true data distribution, reducing their effectiveness. This paper introduces a reliable inference attentive neural process (RIANP) based on normalizing flows (NFs) and neural ordinary differential equations (NODEs), a method for detecting anomalies in industrial equipment. NFs replace the fixed prior assumption of the attentive neural process (ANP) with a learnable prior distribution, addressing sampling holes caused by unlearnable priors. Next, NODEs model the posterior distribution, enabling smoother alignment between the learnable prior and complex posterior distributions. A validation of two industrial anomaly detection cases shows that the RIANP achieves an average F1 score of 94.64 %, a 7.5 % improvement over the ANP, and an AUC of 96.5 %, representing a 12 % enhancement.</div></div>","PeriodicalId":55219,"journal":{"name":"Computers in Industry","volume":"171 ","pages":"Article 104331"},"PeriodicalIF":8.2,"publicationDate":"2025-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144563238","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}

引用次数: 0