查看更多>>摘要：? 2021 Elsevier B.V.X-ray images are commonly used to ensure the security of devices in quality inspection industry. The recognition of signs printed on X-ray weld images plays an essential role in digital traceability system of manufacturing industry. However, the scales of objects vary different greatly in weld images, and it hinders us to achieve satisfactory recognition. In this paper, we propose a signs recognition framework based on convolutional neural networks (CNNs) for weld images. The proposed framework firstly contains a shallow classification network for correcting the pose of images. Moreover, we present a novel spatial and channel enhancement (SCE) module to address the above scale problem. This module can integrate multi-scale features and adaptively assign weights for each feature source. Based on SCE module, a narrow network is designed for final weld information recognition. To enhance the practicability of our framework, we carefully design the architecture of framework with a few parameters and computations. Experimental results show that our framework achieves 99.7% accuracy with 1.1 giga floating-point of operations (GFLOPs) on classification stage, and 90.0 mean average precision (mAP) with 176.1 frames per second (FPS) on recognition stage.
查看更多>>摘要：? 2021 Elsevier B.V.In recent years, the interest in miniaturization of devices easier to wear or even to insert into the body and in general the interest in building lightweight and compact systems are greatly increased. For this reason, the field of micro manufacturing is becoming more and more strategic for modern industry. One critical aspect for the adoption of micro manufacturing technologies on large scale is the quality validation and metrology of components or devices with conventional technologies, such as vision systems or tactile profilometer. Indeed, these technologies hardly fit to be used on the micro devices due to their extremely reduced dimensions which cause a high risk of damaging the micro structured surface. For these reasons, one of the challenges in the micro manufacturing field is the study of methods and tools for the continuous monitoring of the micro production process, with the final aim to improve reliability. In this regard, this paper focuses on a typical micro manufacturing technology, such as micro injection moulding, and presents a methodological approach that can help companies to adopt a solution for optimizing their process leveraging the corresponding Digital Twin. The latter represents a mirror of the physical process that allows to monitor in-line its parameters, to compare them with any analytic model, and to supply specific variations of parameters to keep them always in optimal conditions. The paper also presents a proof of concept of the proposed approach that has been validated to prove the correctness and its capability to scale to a real case study.
查看更多>>摘要：? 2021 Elsevier B.V.A unified view of the Information System (IS) design is essential for dealing with complexity. However, the literature proposes many denominations, depending on the layer, methodology, framework, or tool. This multitude of approaches does not allow a holistic view of the system. Besides, in Information Systems, the search for good practices in design is still a relevant issue. A subset of essential Unified Modeling Language (UML) diagrams is chosen to create a broad view of the IS. CRUD matrix is one of the preferred approaches to articulate the sub-systems of applications and data. Axiomatic Design (AD) provides rules for the improvement of the IS design. This work presents a method to create object-oriented elements based on the CRUD matrix aligned with the business strategy. An integrated student-based case study on logistics is provided. In the discussion, a new IS architect role is proposed supported by the CRUD/AD method.
查看更多>>摘要：? 2021 Elsevier B.V.The architecture, engineering, and construction (AEC) industry has seen a continuous growth in the application of Building Information Modelling (BIM) over the last two decades while building automation and robotics is also gaining ground in the meantime. With more joint applications of both BIM and robotics being exposed, the compatibility of BIM for robotic building construction and maintenance has been found to be fundamentally key to further practical implementations. This paper presents an integrated review combining quantitative and qualitative analyses to clarify the development roadmap of BIM-enabled building automation and robotics. Firstly, information and data management in Digital Twin and robotic building tasks is recapped as the prior knowledge. Secondly, a systematic review on BIM-related research topics in the last 18 years is conducted to demonstrate the inevitability and feasibility of BIM-robot integration, informed by the evolution of lifecycle building information flow. Then, critical analysis on three typical examples of BIM-based robotic construction (i.e. printing, assembling, and welding) are successively performed to reflect the potential BIM-robot integration mechanism and challenges ahead. Finally, a new unified robot-oriented BIM framework, namely BIMfAR, is proposed to conclude the cogitation based on the above analyses. Under the BIMfAR framework, definitions of basic concepts and a practical workflow are discussed to provide guidelines for future applications.
查看更多>>摘要：? 2021 The Author(s)Current trends, such as individualization, increasing complexity, and specialization, require digitization in engineering and production. However, digitization by itself often leads to so-called data silos, which cannot be leveraged effectively when designing and operating smart factories due to the heterogeneity of the information available. This paper presents a framework for (semi-)automatically merging the highly reusable terminological components of production ontologies in an a posteriori way. The framework combines translations, domain-specific vocabularies, and inconsistency checks with syntactic, terminological, and structural analyses to integrate knowledge representations formalized in the Web Ontology Language. Integrating heterogeneous knowledge representations in the production domain can improve support systems for engineers, increase awareness of interdependencies, and enable unambiguous communication in smart factories.
查看更多>>摘要：? 2021 Elsevier B.V.Virtual metrology (VM) is a promising solution for wafer-to-wafer quality monitoring in the semiconductor manufacturing process. VM alternates physical metrology with a prediction model trained using previous metrology data. Active learning can be used to build a VM model for new equipment efficiently with reduced metrology costs. However, conventional active learning is limited by a low prediction accuracy at its initial stage, which is referred to as the cold-start problem. In this study, we propose a domain-adaptive active learning method to address this issue. Using existing equipment as the source domain, the proposed method initializes the VM model through unsupervised domain adaptation from the source domain to the target domain. Active learning is then performed to iteratively update the VM model toward improving the prediction accuracy. Thus, the metrology cost required to obtain a VM model that satisfies the desired prediction accuracy for the target metrology task can be reduced. The effectiveness of the proposed method is demonstrated experimentally using real-world data from a semiconductor manufacturer.
查看更多>>摘要：? 2021 Elsevier B.V.Traditionally, construction managers were considered more experienced in project management than off-site production management, although the latter is gaining importance with the renaissance of modular construction worldwide. Various Internet of Things (IoT)-enabled Building Information Modelling (BIM) platforms have been developed to facilitate production management by providing better information visibility, traceability, and a more collaborative working environment. Nevertheless, by and large, existing platforms suffer from two shortcomings: (a) the ‘single point of failure’ problem of IoT networks and (b) how to guarantee the provenance of BIM modifications from multi-sources. This research aims to develop a blockchain-enabled IoT-BIM platform (BIBP) for off-site production management in modular construction (OPM-MC) that can overcome the shortcomings. A design science research method is adopted to develop a three-layer BIBP system architecture. The system architecture is implemented and then compared with the existing IoT-enabled BIM platform. It was found that BIBP can avoid a single point of failure in IoT networks and ensure the provenance of BIM modifications with reduced storage costs in OPM-MC. The system architecture developed in our study can help the industry advance beyond the rhetoric to develop practical blockchain-enabled IoT-BIM applications. Future works are recommended to fine-tune the platform and test and evaluate it in various scenarios.
查看更多>>摘要：? 2021 Elsevier B.V.Following the trend of Industry 4.0, the business model of steel manufacturing is transforming from a historical inwardly focused supplier/customer relationship to one that embraces the wider end-to-end supply chain and improves productivity more holistically. However, the data and information required for supply chain planning and steelmaking process modelling are normally distributed over scattered sources across organisation boundaries and research communities. This leads to a major problem concerning semantic interoperability. To address this issue, this paper introduces a Common Reference Ontology for Steelmaking (CROS). CROS serves as a shared steelmaking resource and capability model that aims to facilitate knowledge modelling, knowledge sharing and information management. In contrast to most of the existing steelmaking ontologies which merely focus on conceptual modelling, our work pays special attention to the real-world implementation and utilisation aspects of CROS. The functionality and usefulness of CROS is evaluated and tested on a real-world condition-based monitoring and maintenance task for cold rolling mills at Tata Steel in the United Kingdom.
查看更多>>摘要：? 2021 Elsevier B.V.“Product life-cycle management” and “supply chain coordination” are integral concepts in today's competitive business environment, each of which alone is of particular importance in the literature. However, the coordination process of channels with short and long life-cycles may require distinct approaches in practice. This study surveys the coordination strategies in product-service supply chains that benefit from a long product life-cycle. One of the distinguishing features of such channels is that at the start of each life-cycle phase, there is sufficient time to adapt strategies, decisions, and coordination processes to various characteristics of that stage. Coordination of members of product-service supply chain in which the market demand, pricing, and sales effort patterns are dynamically specified following the features of each phase is the general discussion of current research. Three distinct incentive schemes are developed under the Stackelberg game to realize the coordination goals with the use of decentralized and centralized decision-making structures. The analytical results indicate that at the beginning of each stage, the adopted coordination plan should be customized according to the properties of the respective phase because all decisions are varying by passing through the different phases.
查看更多>>摘要：? 2021 Elsevier B.V.As the most important welding method in the welding process of automobile body-in-whites, resistance spot welding still relies on manual inspection of appearance quality to eliminate unqualified products, which leads to a low detection efficiency and a high error rate. In response to this problem, the modular design method is adopted in this paper and a multiscale convolution assemble block is proposed firstly, which can effectively distinguish the differences in similar welding spots. Then a dual-use attention block is designed to calibrate the spatial and channel information of welding spot feature maps, so that the model can pay more attention to the valid welding spot features. Several such multiscale blocks with attention mechanism are stacked to generate the main body of the convolutional neural network (CNN) for welding spot appearance recognition. Finally, the de-pooling and feature fusion strategies are used to further improve the computational efficiency and precision, and the model is named AcmNet. The results of classification experiments demonstrated that the proposed strategies are efficient and the accuracy of AcmNet is 95.2%, which is higher than the existing models and very suitable for the body-in-white production lines.
NSTL
Elsevier