Pub Date : 2023-04-25DOI: 10.48550/arXiv.2304.14275
Peter Meltzer, J. Lambourne, Daniele Grandi
Semantic knowledge of part-part and part-whole relationships in assemblies is useful for a variety of tasks from searching design repositories to the construction of engineering knowledge bases. In this work we propose that the natural language names designers use in Computer Aided Design (CAD) software are a valuable source of such knowledge, and that Large Language Models (LLMs) contain useful domain-specific information for working with this data as well as other CAD and engineering-related tasks. In particular we extract and clean a large corpus of natural language part, feature and document names and use this to quantitatively demonstrate that a pre-trained language model can outperform numerous benchmarks on three self-supervised tasks, without ever having seen this data before. Moreover, we show that fine-tuning on the text data corpus further boosts the performance on all tasks, thus demonstrating the value of the text data which until now has been largely ignored. We also identify key limitations to using LLMs with text data alone, and our findings provide a strong motivation for further work into multi-modal text-geometry models. To aid and encourage further work in this area we make all our data and code publicly available.
{"title":"What's in a Name? Evaluating Assembly-Part Semantic Knowledge in Language Models through User-Provided Names in CAD Files","authors":"Peter Meltzer, J. Lambourne, Daniele Grandi","doi":"10.48550/arXiv.2304.14275","DOIUrl":"https://doi.org/10.48550/arXiv.2304.14275","url":null,"abstract":"\u0000 Semantic knowledge of part-part and part-whole relationships in assemblies is useful for a variety of tasks from searching design repositories to the construction of engineering knowledge bases. In this work we propose that the natural language names designers use in Computer Aided Design (CAD) software are a valuable source of such knowledge, and that Large Language Models (LLMs) contain useful domain-specific information for working with this data as well as other CAD and engineering-related tasks. In particular we extract and clean a large corpus of natural language part, feature and document names and use this to quantitatively demonstrate that a pre-trained language model can outperform numerous benchmarks on three self-supervised tasks, without ever having seen this data before. Moreover, we show that fine-tuning on the text data corpus further boosts the performance on all tasks, thus demonstrating the value of the text data which until now has been largely ignored. We also identify key limitations to using LLMs with text data alone, and our findings provide a strong motivation for further work into multi-modal text-geometry models. To aid and encourage further work in this area we make all our data and code publicly available.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":3.1,"publicationDate":"2023-04-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85029106","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Bingkun Yuan, Junnan Ye, Xinying Wu, Chaoxiang Yang
With the development of social productivity and the improvement in material living standards, emotional value has become the core driver of the enhancement of product market competitiveness. A medical nursing bed, one of the most typical types of medical devices, is designed with little attention to the emotional experience of the users. Therefore, this paper proposes an innovative perceptual design approach under the Kansei engineering (KE) framework for resource-limited and information-poor companies. It guides the aesthetic design of medical nursing beds by constructing a mapping relationship between users' perceptual needs and the design characteristics of medical nursing beds to maximize users' emotions. First, latent Dirichlet allocation (LDA) is used to extract usable Kansei semantics from big data, compensating for the subjectivity of traditional KE data input. Then, the design characteristics obtained after deconstructing a medical nursing bed are simplified with rough set theory (RST). Finally, a mapping model between users' perceptual needs and the core design characteristics of nursing beds is established through support vector regression (SVR), and the optimal design solution is obtained by weighting calculation. The optimal combination of design characteristics for medical nursing beds is finally obtained. The results suggest that the design method proposed in this paper can help designers accurately grasp users' emotional perceptions in terms of aesthetic design and scientifically guide and complete the design of new medical nursing beds, verifying the feasibility and scientificity of the proposed method in terms of aesthetic design.
{"title":"Applying Latent Dirichlet Allocation and Support Vector Regression to the Aesthetic Design of Medical Nursing Beds","authors":"Bingkun Yuan, Junnan Ye, Xinying Wu, Chaoxiang Yang","doi":"10.1115/1.4062350","DOIUrl":"https://doi.org/10.1115/1.4062350","url":null,"abstract":"\u0000 With the development of social productivity and the improvement in material living standards, emotional value has become the core driver of the enhancement of product market competitiveness. A medical nursing bed, one of the most typical types of medical devices, is designed with little attention to the emotional experience of the users. Therefore, this paper proposes an innovative perceptual design approach under the Kansei engineering (KE) framework for resource-limited and information-poor companies. It guides the aesthetic design of medical nursing beds by constructing a mapping relationship between users' perceptual needs and the design characteristics of medical nursing beds to maximize users' emotions. First, latent Dirichlet allocation (LDA) is used to extract usable Kansei semantics from big data, compensating for the subjectivity of traditional KE data input. Then, the design characteristics obtained after deconstructing a medical nursing bed are simplified with rough set theory (RST). Finally, a mapping model between users' perceptual needs and the core design characteristics of nursing beds is established through support vector regression (SVR), and the optimal design solution is obtained by weighting calculation. The optimal combination of design characteristics for medical nursing beds is finally obtained. The results suggest that the design method proposed in this paper can help designers accurately grasp users' emotional perceptions in terms of aesthetic design and scientifically guide and complete the design of new medical nursing beds, verifying the feasibility and scientificity of the proposed method in terms of aesthetic design.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":3.1,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89095496","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abheek Chatterjee, Cade Helbig, Richard Malak, Astrid Layton
Abstract System of systems (SoS) are networked integration of constituent systems that together achieve new capabilities not possible through the operation of any single system. SoS can be found across all aspects of modern life such as power grids, supply chains, and disaster monitoring and tracking services. Their resilience (being able to withstand and recover from disruptions) is a critical attribute whose evaluation is nontrivial and requires detailed disruption models. Engineers rely on heuristics (such as redundancy and localized capacity) for achieving resilience. However, excessive reliance on these qualitative guidelines can result in unacceptable operation costs, erosion of profits, over-consumption of natural resources, or unacceptable levels of waste or emissions. Graph-theoretic approaches provide a potential solution to this challenge as they can evaluate architectural characteristics without needing detailed performance simulations, supporting their use in early stage SoS architecture selection. However, no consensus exists as to which graph-theoretic metrics are most valuable for SoS design and how they should be included in the design process. In this work, multiple graph-theoretic approaches are analyzed and compared, on a common platform, for their use as design tools for resilient SoS. The metrics central point dominance, modularity, specialized predator ratio, generalization, vulnerability, and degree of system order are found to be viable options for the development of early stage decision-support tools for resilient SoS design.
{"title":"A Comparison of Graph-Theoretic Approaches for Resilient System of Systems Design","authors":"Abheek Chatterjee, Cade Helbig, Richard Malak, Astrid Layton","doi":"10.1115/1.4062231","DOIUrl":"https://doi.org/10.1115/1.4062231","url":null,"abstract":"Abstract System of systems (SoS) are networked integration of constituent systems that together achieve new capabilities not possible through the operation of any single system. SoS can be found across all aspects of modern life such as power grids, supply chains, and disaster monitoring and tracking services. Their resilience (being able to withstand and recover from disruptions) is a critical attribute whose evaluation is nontrivial and requires detailed disruption models. Engineers rely on heuristics (such as redundancy and localized capacity) for achieving resilience. However, excessive reliance on these qualitative guidelines can result in unacceptable operation costs, erosion of profits, over-consumption of natural resources, or unacceptable levels of waste or emissions. Graph-theoretic approaches provide a potential solution to this challenge as they can evaluate architectural characteristics without needing detailed performance simulations, supporting their use in early stage SoS architecture selection. However, no consensus exists as to which graph-theoretic metrics are most valuable for SoS design and how they should be included in the design process. In this work, multiple graph-theoretic approaches are analyzed and compared, on a common platform, for their use as design tools for resilient SoS. The metrics central point dominance, modularity, specialized predator ratio, generalization, vulnerability, and degree of system order are found to be viable options for the development of early stage decision-support tools for resilient SoS design.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135708616","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Practical manufacturing system operates in highly dynamic and uncertain environments, where stochastic disturbances disrupt the execution of production schedule as originally developed. Previous dynamic scheduling mainly focuses on the constructing predictive models for machine unavailability, with little studies on the adaptive and self-learning capacities for changing scheduling environments. Therefore, a digital twin (DT) driven scheduling with dynamic feedback mechanism is proposed, in which a reinforcement learning (RL) based adaptive scheduling is developed in DT to make corrective decisions for the disturbances during production runs. In the proposed architecture, the happening disturbance is firstly detected in the virtual layer by the status continuously updating in accordance with the physical workshop. Furthermore, the reschedule triggering condition is determined in real-time through quantization of the progress deviations resulting from disturbances. For the scheduling approach, the multi-agents RL (MARL) based adaptive scheduling method is built to perceive the dynamic production status from virtual environment and implement corrective strategies to hedge against the occurred disturbances. Finally, the proposed method is verified by a practical job shop case and the corresponding DT system is developed to show the effectiveness and advantages after a practical implementation.
{"title":"An Adaptive Job Shop Scheduling Mechanism for Disturbances by Running Reinforcement Learning in Digital Twin Environment","authors":"Weiguang Fang, Hao Zhang, Weiwei Qian, Yuhao Guo, Shaoxun Li, Zeqing Liu, Chenning Liu, Dongpao Hong","doi":"10.1115/1.4062349","DOIUrl":"https://doi.org/10.1115/1.4062349","url":null,"abstract":"\u0000 Practical manufacturing system operates in highly dynamic and uncertain environments, where stochastic disturbances disrupt the execution of production schedule as originally developed. Previous dynamic scheduling mainly focuses on the constructing predictive models for machine unavailability, with little studies on the adaptive and self-learning capacities for changing scheduling environments. Therefore, a digital twin (DT) driven scheduling with dynamic feedback mechanism is proposed, in which a reinforcement learning (RL) based adaptive scheduling is developed in DT to make corrective decisions for the disturbances during production runs. In the proposed architecture, the happening disturbance is firstly detected in the virtual layer by the status continuously updating in accordance with the physical workshop. Furthermore, the reschedule triggering condition is determined in real-time through quantization of the progress deviations resulting from disturbances. For the scheduling approach, the multi-agents RL (MARL) based adaptive scheduling method is built to perceive the dynamic production status from virtual environment and implement corrective strategies to hedge against the occurred disturbances. Finally, the proposed method is verified by a practical job shop case and the corresponding DT system is developed to show the effectiveness and advantages after a practical implementation.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":3.1,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83762917","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Conceptualization and design of intellectualized, socialized, and personalized cyber-physical systems (CPSs) needs exploration and synthesis of novel knowledge. In turn, it raises the need for a combined use of interdisciplinary, multidisciplinary, and transdisciplinary research. Supradisciplinary research emerged as a new doctrine of combining these research approaches from epistemological, methodological, and procedural perspective. However, no methodology can be found in the literature that could facilitate the practical execution of supradisciplinary research programs and projects. This position paper proposes a conceptual framework that can be used as a blueprint of operationalization of such undertakings. The framework rests on six generic pillars: (i) problematics, (ii) infrastructure, (iii) methodics, (iv) stakeholders, (v) operations, and (vi) knowledge.The framework arranges the concerns in a procedural logic - as they should be considered by the research managers and cyber-physical system developers. In its current form, the framework does not cover the specific societal and personal issues of a successful organization of the inquiry at individual researchers, research teams, and research community levels. Notwithstanding, the framework can facilitate management of research organization tasks, joint formation of shared research infrastructure, setting up concrete research programs, projects, and processes, academic partnering and public stakeholder involvement, process flow management and capacity/competence allocation, and knowledge synthesis, assessment, and consolidation in a holistic manner. Follow up community-based research may focus on the practical application and testing of the framework in concrete cases – a task that an individual researcher cannot address.
{"title":"Framing Supradisciplinary Research for Intellectualized Cyber-Physical Systems: An Unfinished Story","authors":"I. Horváth","doi":"10.1115/1.4062327","DOIUrl":"https://doi.org/10.1115/1.4062327","url":null,"abstract":"\u0000 Conceptualization and design of intellectualized, socialized, and personalized cyber-physical systems (CPSs) needs exploration and synthesis of novel knowledge. In turn, it raises the need for a combined use of interdisciplinary, multidisciplinary, and transdisciplinary research. Supradisciplinary research emerged as a new doctrine of combining these research approaches from epistemological, methodological, and procedural perspective. However, no methodology can be found in the literature that could facilitate the practical execution of supradisciplinary research programs and projects. This position paper proposes a conceptual framework that can be used as a blueprint of operationalization of such undertakings. The framework rests on six generic pillars: (i) problematics, (ii) infrastructure, (iii) methodics, (iv) stakeholders, (v) operations, and (vi) knowledge.The framework arranges the concerns in a procedural logic - as they should be considered by the research managers and cyber-physical system developers. In its current form, the framework does not cover the specific societal and personal issues of a successful organization of the inquiry at individual researchers, research teams, and research community levels. Notwithstanding, the framework can facilitate management of research organization tasks, joint formation of shared research infrastructure, setting up concrete research programs, projects, and processes, academic partnering and public stakeholder involvement, process flow management and capacity/competence allocation, and knowledge synthesis, assessment, and consolidation in a holistic manner. Follow up community-based research may focus on the practical application and testing of the framework in concrete cases – a task that an individual researcher cannot address.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":3.1,"publicationDate":"2023-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79508965","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Thakur, Swagatika Sahoo, Arnab Mukherjee, Raju Halder
Lately the importance of swarm robotics has been recognized in a wide range of areas, including logistics, surveillance, disaster management, agriculture, and other industrial applications. The swarm intelligence introduced by the existing paradigm of Artificial Intelligence and Machine Learning often ignores the aspect of providing security and reliability guarantees. Consider a futuristic scenario wherein self-driving cars will transport people, self-driving trucks will carry cargo between warehouses, and a combination of legged robots/drones will ship cargo from warehouses to doorsteps. In the case of such a heterogeneous swarm of robots, it is crucial to ensure a trustful and reliable operating platform for smooth coordination, collaborative decision-making via appropriate consensus, and seamless information sharing while ensuring data security. In this direction, blockchain has been proven to be an effective technology that maintains the transactions (records) in a trustful manner after being validated through consensus. This guarantees accountability, transparency, and trust concerning the storage, safeguarding, and sharing of information among the parties. In this paper, we provide a walkthrough demonstrating the feasibility of using blockchain technology to make the robotic swarm trustful systems in their adoption to critical applications at large-scale. We highlight the pros and cons of the use of cloud vis-a-vis blockchain in swarm robotics. Finally, we present various future research opportunities pertaining to the adoption of blockchain technology in swarm robotics applications.
{"title":"Making Robotic Swarms Trustful: A Blockchain-Based Perspective","authors":"A. Thakur, Swagatika Sahoo, Arnab Mukherjee, Raju Halder","doi":"10.1115/1.4062326","DOIUrl":"https://doi.org/10.1115/1.4062326","url":null,"abstract":"\u0000 Lately the importance of swarm robotics has been recognized in a wide range of areas, including logistics, surveillance, disaster management, agriculture, and other industrial applications. The swarm intelligence introduced by the existing paradigm of Artificial Intelligence and Machine Learning often ignores the aspect of providing security and reliability guarantees. Consider a futuristic scenario wherein self-driving cars will transport people, self-driving trucks will carry cargo between warehouses, and a combination of legged robots/drones will ship cargo from warehouses to doorsteps. In the case of such a heterogeneous swarm of robots, it is crucial to ensure a trustful and reliable operating platform for smooth coordination, collaborative decision-making via appropriate consensus, and seamless information sharing while ensuring data security. In this direction, blockchain has been proven to be an effective technology that maintains the transactions (records) in a trustful manner after being validated through consensus. This guarantees accountability, transparency, and trust concerning the storage, safeguarding, and sharing of information among the parties. In this paper, we provide a walkthrough demonstrating the feasibility of using blockchain technology to make the robotic swarm trustful systems in their adoption to critical applications at large-scale. We highlight the pros and cons of the use of cloud vis-a-vis blockchain in swarm robotics. Finally, we present various future research opportunities pertaining to the adoption of blockchain technology in swarm robotics applications.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":3.1,"publicationDate":"2023-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80189293","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
W. Bernstein, Andrew Bowman, R. Durscher, A. Gillman, S. Donegan
Extended reality (XR) technologies have realized significant value for design, manufacturing, and sustainment processes. However, Industrial XR, or XR implemented within industrial applications, suffers from scalability and flexibility challenges due to fundamental gaps with interoperability between data, models, and platforms. Though there has been a number of recent efforts to improve the interoperability of industrial XR technologies, progress has been hindered by an innate separation between the domain-specific models (e.g., manufacturing execution data, material specifications, and product manufacturing information) with XR (often-standard) processes (e.g., multi-scale spatial representations and data formats optimized for run-time presentation). In this paper, we elaborate on promising research directions and opportunities around which the manufacturing and visualization academic community can rally. To establish such research directions, we (1) conducted a meta-review on well-established state-of-the-art review articles that have already presented in-depth surveys on application areas for industrial XR, such as maintenance, assembly and inspection and (2) mapped those findings to publicly published priorities from across the US Department of Defense. We hope that our presented research agenda will spur interdisciplinary work across academic silos, i.e., manufacturing and visualization communities, and engages either community within work groups led by the other, e.g., within standards development organizations.
{"title":"Towards Data and Model Interoperability for Industrial Extended Reality in Manufacturing","authors":"W. Bernstein, Andrew Bowman, R. Durscher, A. Gillman, S. Donegan","doi":"10.1115/1.4062328","DOIUrl":"https://doi.org/10.1115/1.4062328","url":null,"abstract":"\u0000 Extended reality (XR) technologies have realized significant value for design, manufacturing, and sustainment processes. However, Industrial XR, or XR implemented within industrial applications, suffers from scalability and flexibility challenges due to fundamental gaps with interoperability between data, models, and platforms. Though there has been a number of recent efforts to improve the interoperability of industrial XR technologies, progress has been hindered by an innate separation between the domain-specific models (e.g., manufacturing execution data, material specifications, and product manufacturing information) with XR (often-standard) processes (e.g., multi-scale spatial representations and data formats optimized for run-time presentation). In this paper, we elaborate on promising research directions and opportunities around which the manufacturing and visualization academic community can rally. To establish such research directions, we (1) conducted a meta-review on well-established state-of-the-art review articles that have already presented in-depth surveys on application areas for industrial XR, such as maintenance, assembly and inspection and (2) mapped those findings to publicly published priorities from across the US Department of Defense. We hope that our presented research agenda will spur interdisciplinary work across academic silos, i.e., manufacturing and visualization communities, and engages either community within work groups led by the other, e.g., within standards development organizations.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":3.1,"publicationDate":"2023-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87669241","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Megacasting is a new concept in the automotive industry. A large number of sheet metal parts will be replaced with one large aluminum casting, i.e. a megacasting. This helps to reduce weight, opens up for larger design flexibility, allows for a more circular production, and takes away a large number of assembly steps in the production process. However, there are also challenges related to the use of megacastings. This position paper outlines challenges associated with the geometrical quality of the final product. It covers robust design and tolerancing in early product development phases as well as inspection preparation during pre-production and digital twin set-up during full production to ensure the geometrical quality of a product containing a megacasting. Simulations of both part level and assembly level deviation and variation are discussed. The paper outlines a geometry assurance process for products containing megacastings in the automotive industry, and what research challenges that are the most important ones to address in this area. It is concluded that computer-aided tolerancing tools must be able to predict the dimensional effects from joining methods such as flow drill fasteners or self-pierced riveting, to use casting simulation as input, and to handle combinations of solid and surface meshes. Furthermore, there might be a need for adjustments to the joining process based on digital twins to achieve proper quality at a reasonable price. Experiences in using megacastings in the body-in-white are lacking and a fast learning curve is required.
{"title":"Challenges in Geometry Assurance of Megacasting in the Automotive Industry","authors":"Kristina Wärmefjord, Josefin Hansen, R. Söderberg","doi":"10.1115/1.4062269","DOIUrl":"https://doi.org/10.1115/1.4062269","url":null,"abstract":"\u0000 Megacasting is a new concept in the automotive industry. A large number of sheet metal parts will be replaced with one large aluminum casting, i.e. a megacasting. This helps to reduce weight, opens up for larger design flexibility, allows for a more circular production, and takes away a large number of assembly steps in the production process. However, there are also challenges related to the use of megacastings. This position paper outlines challenges associated with the geometrical quality of the final product. It covers robust design and tolerancing in early product development phases as well as inspection preparation during pre-production and digital twin set-up during full production to ensure the geometrical quality of a product containing a megacasting. Simulations of both part level and assembly level deviation and variation are discussed. The paper outlines a geometry assurance process for products containing megacastings in the automotive industry, and what research challenges that are the most important ones to address in this area. It is concluded that computer-aided tolerancing tools must be able to predict the dimensional effects from joining methods such as flow drill fasteners or self-pierced riveting, to use casting simulation as input, and to handle combinations of solid and surface meshes. Furthermore, there might be a need for adjustments to the joining process based on digital twins to achieve proper quality at a reasonable price. Experiences in using megacastings in the body-in-white are lacking and a fast learning curve is required.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":3.1,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73068823","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract Given a part design, the task of manufacturing process classification identifies an appropriate manufacturing process to fabricate it. Our previous research proposed a large dataset for manufacturing process classification and achieved accurate classification results based on a combination of a convolutional neural network (CNN) and the heat kernel signature for triangle meshes. In this paper, we constructed a classification method based on rotation invariant shape descriptors and a neural network, and it achieved better accuracy than all previous methods. This method uses a point cloud part representation, in contrast to the triangle mesh representation used in our previous work. The first step extracted rotation invariant features consisting of a set of distances between points in the point cloud. Then, the extracted shape descriptors were fed into a CNN for the classification of manufacturing processes. In addition, we provide two visualization methods for interpreting the intermediate layers of the neural network. Last, the performance of the method was tested on some ambiguous examples and their performances were consistent with expectations. In this paper, we have considered only shape information, while non-shape information like materials and tolerances were ignored. Additionally, only parts that require one manufacturing process were considered in this research. Our work demonstrates that part shape attributes alone are adequate for discriminating between different manufacturing processes considered.
{"title":"Manufacturing Process Classification Based on Distance Rotationally Invariant Convolutions","authors":"Zhichao Wang, David Rosen","doi":"10.1115/1.4056806","DOIUrl":"https://doi.org/10.1115/1.4056806","url":null,"abstract":"Abstract Given a part design, the task of manufacturing process classification identifies an appropriate manufacturing process to fabricate it. Our previous research proposed a large dataset for manufacturing process classification and achieved accurate classification results based on a combination of a convolutional neural network (CNN) and the heat kernel signature for triangle meshes. In this paper, we constructed a classification method based on rotation invariant shape descriptors and a neural network, and it achieved better accuracy than all previous methods. This method uses a point cloud part representation, in contrast to the triangle mesh representation used in our previous work. The first step extracted rotation invariant features consisting of a set of distances between points in the point cloud. Then, the extracted shape descriptors were fed into a CNN for the classification of manufacturing processes. In addition, we provide two visualization methods for interpreting the intermediate layers of the neural network. Last, the performance of the method was tested on some ambiguous examples and their performances were consistent with expectations. In this paper, we have considered only shape information, while non-shape information like materials and tolerances were ignored. Additionally, only parts that require one manufacturing process were considered in this research. Our work demonstrates that part shape attributes alone are adequate for discriminating between different manufacturing processes considered.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135469256","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Wentai Zhang, Joe Joseph, Quan Chen, Can Koz, Liuyue Xie, Amit Regmi, Soji Yamakawa, T. Furuhata, Kenji Shimada, L. Kara
We present a new data generation method to facilitate an automatic machine interpretation of 2D engineering part drawings. While such drawings are a common medium for clients to encode design and manufacturing requirements, a lack of computer support to automatically interpret these drawings necessitates part manufacturers to resort to laborious manual approaches for interpretation which, in turn, severely limits processing capacity. Although recent advances in trainable computer vision methods may enable automatic machine interpretation, it remains challenging to apply such methods to engineering drawings due to a lack of labeled training data. As one step toward this challenge, we propose a constrained data synthesis method to generate an arbitrarily large set of synthetic training drawings using only a handful of labeled examples. Our method is based on the randomization of the dimension sets subject to two major constraints to ensure the validity of the synthetic drawings. The effectiveness of our method is demonstrated in the context of a binary component segmentation task with a proposed list of descriptors. An evaluation of several image segmentation methods trained on our synthetic dataset shows that our approach to new data generation can boost the segmentation accuracy and the generalizability of the machine learning models to unseen drawings.
{"title":"A Data Augmentation Method for Data-Driven Component Segmentation of Engineering Drawings","authors":"Wentai Zhang, Joe Joseph, Quan Chen, Can Koz, Liuyue Xie, Amit Regmi, Soji Yamakawa, T. Furuhata, Kenji Shimada, L. Kara","doi":"10.1115/1.4062233","DOIUrl":"https://doi.org/10.1115/1.4062233","url":null,"abstract":"\u0000 We present a new data generation method to facilitate an automatic machine interpretation of 2D engineering part drawings. While such drawings are a common medium for clients to encode design and manufacturing requirements, a lack of computer support to automatically interpret these drawings necessitates part manufacturers to resort to laborious manual approaches for interpretation which, in turn, severely limits processing capacity. Although recent advances in trainable computer vision methods may enable automatic machine interpretation, it remains challenging to apply such methods to engineering drawings due to a lack of labeled training data. As one step toward this challenge, we propose a constrained data synthesis method to generate an arbitrarily large set of synthetic training drawings using only a handful of labeled examples. Our method is based on the randomization of the dimension sets subject to two major constraints to ensure the validity of the synthetic drawings. The effectiveness of our method is demonstrated in the context of a binary component segmentation task with a proposed list of descriptors. An evaluation of several image segmentation methods trained on our synthetic dataset shows that our approach to new data generation can boost the segmentation accuracy and the generalizability of the machine learning models to unseen drawings.","PeriodicalId":54856,"journal":{"name":"Journal of Computing and Information Science in Engineering","volume":null,"pages":null},"PeriodicalIF":3.1,"publicationDate":"2023-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81558278","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：