查看更多>>摘要：Digital twins present a conceptual framework for product life-cycle monitoring and control using a simulated replica of the physical system. Since their emergence, they have garnered particular attention as a shift away from costly physical testing and towards the use of high fidelity simulations, sensor data and intelligent control. Metal additive manufacturing (AM), a 3D printing technology prone to defects, requires a digital twin capable of tackling issues of printed part qualification, certification and optimisation. In this paper, we evaluate the key features specific to metal AM and review the current literature of modelling, sensing, control and machine intelligence. We find that the body of research toward the development of an metal additive manufacturing (AM) digital twin can be organised logically into a hierarchy of four levels of increasing complexity. The elements composing each level require deep integration and we highlight the key enabling technologies: surrogate modelling, in-situ sensing, hardware control systems and intelligent control policies. Our proposed digital twin hierarchy for AM provides a developer framework for engineering digital twins, both for AM and other intelligent manufacturing systems.(c) 2022 Published by Elsevier B.V.
查看更多>>摘要：Industrial Internet of Things (I-IoT) is a network of devices that focus on monitoring industrial assets and continuously collecting data. This data can be utilized by Machine Learning (ML) methods to perform Predictive Maintenance (PDM) which identifies an optimal maintenance schedule for the industrial assets. The computational systems in the I-IoT are usually not designed with security in mind. Their limited computational power creates security vulnerabilities that attackers can exploit to prevent asset availability, sabotage communication, and corrupt system data. In this work, we first demonstrate that cyber-attacks can impact the performance of ML-based PDM methods significantly, leading up to 120 x prediction performance loss. Next, we develop a stacking ensemble learning-based framework that stays resilient against various white-box adversarial attacks. The results show that our framework performs well in the presence of cyber-attacks and has up to 60% higher resiliency compared to the most resilient individual ML method.
查看更多>>摘要：Industry 4.0 (I4.0) has brought several challenges related to the need to acquire and integrate large amounts of data from multiple sources in order to integrate these elements into an automated manufacturing system. Establishing interoperability is crucial to meet these challenges, and standards development and adoptions play a key role in achieving this. Therefore, academics and industrial stakeholders must join their forces in order to develop methods to enhance interoperability and to mitigate possible conflicts between standards. The aim of this paper is to propose an approach that enhances interoperability between standards through the combined use of Natural Language Processing (NLP) and Knowledge Graphs (KG). In particular, the proposed method is based on the measurement of semantic similarity among the textual content of standards documents belonging to different standardization frameworks. The present study contributes to the research and practice in three ways. First, it fills research gaps concerning the synergy of NLP, KGs and I4.0. Second, it provides an automatic method that improves the process of creating, curating and enriching a KG. Third, it provides qualitative and quantitative evidence of Semantic Interoperability Conflicts (SICs). The experimental results of the application of the proposed method to the I4.0 Standards Knowledge Graph (I40KG) show that different standards are still struggling to use a shared language and that there exists a strong different in the view of I4.0 proposed by the two main standardization frameworks (RAMI and IIRA). By automatically enriching the I40KG with a solid experimental approach, we are paving the way for actionable knowledge which has been extracted from the PDFs and made available in the I40KG.(c) 2022 Elsevier B.V. All rights reserved.
查看更多>>摘要：Surface defect detection using deep learning approaches has become a promising area of research, but the difficulty of accurately locating and segmenting various forms of defects presents a challenge for this method. Swin Transformer, as a Transformer-based model, has made significant progress in computer vision. Its performance surpasses standard CNN's performance on most tasks, but it has drawn scant attention from industrial applications. Thus far, using CNNs for surface defect detection tends to be the most common application. To explore the extensibility of the Transformer, we seek to expand the applicability of the Swin Transformer and apply it to our task. This paper proposes an improved structure called the Variant Swin Transformer. We designed a new window shift scheme that further strengthens the feature transfer between windows and makes the framework more capable of serving as a backbone for defect detection. The overall framework named the Cas-VSwin Transformer outperformed most existing models on the private dataset we built (82.3 box AP and 80.2 mask AP). We also further verified the superiority of transfer learning in training small-scale datasets. Moreover, the proposed VSwin Transformer has a lower relative error in the quantitative analysis of the defect areas, demonstrating that the Cas-VSwin Transformer is an effective model for surface defect detection, and it has great potential for other similar industrial applications.(c) 2022 Elsevier B.V. All rights reserved.
查看更多>>摘要：For the past few years, we have been hearing about Industry 4.0 (or the fourth industrial revolution), which promises to improve productivity, flexibility, quality, customer satisfaction and employee well-being. To assess whether these goals are achieved, it is necessary to implement a performance management system (PMS). However, a PMS must take into account the various challenges associated with Industry 4.0, including the availability of large amounts of data. While it represents an opportunity for companies to improve performance, big data does not necessarily mean good data. It can be uncertain, imprecise, ambiguous, etc. Uncertainty is one of the major challenges and it is essential to take it into account when computing performance indicators to increase confidence in decision making. To address this issue, we propose a method to model uncertainty in key performance indicators (KPIs). Our work allows associating with each indicator an uncertainty noted m, computed on the basis of the theory of belief functions. The KPI and its associated uncertainty form a pair (KP I, m). The method developed allows calculating this uncertainty m for the input data of the performance management system. We show how these modeled uncertainties should be propagated to the KPIs. For these KPI uncertainties, we have defined rules to support decision-making. The method developed, based on the theory of belief functions, is part of a methodology we propose to define and extract smart data from massive data. To our knowledge, this is the first attempt to use this theory to model uncertain performance indicators. Our work has shown its effectiveness and its applicability to a case of bottle filling line simulation. In addition to these results, this work opens up new perspectives, particularly for taking uncertainty into account in expert opinions and in industrial risk assessment.(c) 2022 Published by Elsevier B.V.
查看更多>>摘要：In the light of Industry 4.0, the field of Industrial Maintenance faces a large digital transformation, adopting Extended Reality (XR) technologies to aid industrial operations. For the manufacturing corporations that provide maintenance services, the efficiency of industrial maintenance plays a crucial role in the competitiveness and is tightly related to the technical documentation supporting maintenance. However, the process of documentation creation faces several challenges due to lack of access to the physical equipment and difficulties in remote communication between globally distributed departments. To address these shortcomings, this research investigates the utilization of Virtual Reality (VR) to facilitate asynchronous collaboration of globally dispersed departments involved in the pipeline of maintenance method and documentation creation. The presented proof-of-concept (the COVE-VR platform) has been developed as an academia-industry collaboration and evaluated iteratively with subject matter experts. The proposed VR platform consists of two virtual environments and eight virtual tools, which allow interaction with virtual prototypes (3D CAD models) and means of digital content creation. Our findings show the high relevance of the developed solution for the needs of industrial departments and the ability to support asynchronous collaboration among them. This article delivers qualitative findings on the value of VR technology and presents guidelines on how to develop virtual tools for digital content creation within VR, adaptable to other industrial contexts. We suggest providing embedded guidance and design consistency to ensure smooth interactions with virtual tools and further discuss the importance of proper positioning, the transparency of operations and the information property of generated content. (c) 2022 The Author(s). Published by Elsevier B.V. CC_BY_4.0
查看更多>>摘要：Currently, computer-aided design (CAD) systems are very important for the design of any engineering products. CAD systems have undergone enormous progress. They are becoming more and more intelligent and automated. However, those systems still need improvement and still have flaws that need to be corrected in order to improve and develop the productivity and efficiency of the design process. Mechanical products are often complex and composed of multiple components so, consequently, an Engineering Change (EC) that affects a module or a single component will certainly influence the rest of the assembly. This phenomenon is called Change Propagation (CP). Therefore, in order to propagate a change that affects an assembly component easily, it is important to predict the Change Propagation Path and the impact of this modification on the rest of the assembly components. Hence, CP becomes an easy task. This paper focuses on the risk assessment of an engineering change, in order to enhance the predictive capacity of engineering change management methods. In fact, a full analysis must be made to measure the real impact of such modifications as well as their feasibility. To this end, the authors proposed a matrix-based approach, which results in a Risk Matrix. This is the product of the Relationships Matrix and the Influences Matrix which are constructed based on the data extracted at an early stage. In fact, the CAD assembly model contains a lot of important information that is extracted using Application Programming Interface (API).(c) 2022 Published by Elsevier B.V.
查看更多>>摘要：Although companies systematically strive for a full digitalisation of their products and their processes, the design phase shows that the quality of models is very unequal. Indeed, detailed design benefits from much more sophisticated methods and tools than the specification and architecture activities. Although, we should note the recent paradigm shift from document-based to model-based systems engineering, these models, which are mainly static 2D diagrams, remain poor to facilitate design verification early on. Thus, to detect most errors during the design phase, companies have no other alternative than to wait up to the testing phase which occurs after several years of development for complex systems. Thus, we propose a user-centric computer-aided verification process to ensure that the design meets the requirements under realistic operational conditions. The verification process provides a progressive immersion into the virtual system before seamlessly transitioning to the real system. Our work is built upon state-of-the-art MBSE methods such as the Property Model Methodology, which enables systems engineers to co-simulate specification models and design models. We improve such MBSE methods by increasing the level of realism that experiences the end-user during the verification of a design by the original combination of Model-In the-Loop, Immersive Model-In-the-Loop, Human-In-the-Loop, and Hardware-In-the-Loop simulation strategies. A robot arm is used as a use case to illustrate the verification process.(c) 2022 Elsevier B.V. All rights reserved.
Wang, RuoxinCheung, Chi FaiWang, ChunjinCheng, Mei Na...
9页
查看更多>>摘要：Surface defects in the selective laser melting (SLM) process adversely affect the surface quality of additive manufacturing workpieces. Although some studies have been conducted to classify or detect defects, the investigation of the distribution of defects has received relatively little attention. Currently, there are many studies which focus on the counting of objects and they typically adopt a backbone convolutional neural network to obtain an initial feature map, which contains more semantic information but loses some geometric details. In this paper, the distribution of surface defects and defect count estimation of additive manufactured components are first studied as well as presentation of a developed deep learning characterization method (DLCM) based on a detail-aware dilated convolutional neural network (DDCNN) incorporated with a fine details feature map extractor designed to obtain final fine semantic features. It features two dilated convolutional layer combination blocks (DCBs), which are proposed to fuse low-level features with semantic features. Data are acquired from the surface of a workpiece fabricated by SLM. A series of experiments have been conducted to validate the performance of the DLCM. Compared with the other main state-of-the-art methods, the proposed DLCM yields better results.(c) 2022 Elsevier B.V. All rights reserved.
查看更多>>摘要：With the emergence of the Industrial Internet of Things (IIoT), potential threats to smart manufacturing systems are increasingly becoming challenging, causing severe damage to production operations and vital industrial assets, even sensitive information. Hence, detecting irregularities for time-series data in industrial control systems that should operate continually is critical, ensuring security and minimizing maintenance costs. In this study, with the hybrid design of Federated learning, Autoencoder, Transformer, and Fourier mixing sublayer, we propose a robust distributed anomaly detection architecture that works more accurately than several most recent anomaly detection solutions within the ICS contexts, whilst being fast learning in minute time scale. This distributed architecture is also proven to achieve lightweight, consume little CPU and memory usage, have low communication costs in terms of bandwidth consumption, which makes it feasible to be deployed on top of edge devices with limited computing capacity.(c) 2022 Elsevier B.V. All rights reserved.
NSTL
Elsevier