IFIP Open Access Digital Library
Not a member yet
22614 research outputs found
Sort by
Privacy Preserving Federated Learning: A Novel Approach for Combining Differential Privacy and Homomorphic Encryption
No full textInternational audienceEnsuring the data security and privacy stands as a prominent concern in the landscape of machine learning. The conventional approach of centralizing training data raises privacy concerns. Federated learning addresses this by avoiding the need to transfer local data when training a global model, opting to share only local model updates. Despite this, the challenge of information leakage persists. Various attempts tried to tackle this issue, but existing solutions lead to a tradeoff between accuracy, privacy and computation time. This is an unevitable challenge. In this paper, we address that challenge by combining differential privacy and homomorphic encryption. This approach allow to add less noise to the data by shuffling to anonymize the data, not only at the client level but also at the parameter level. Hence, it improves the accuracy of the output models while offering strong privacy guarantees. Importantly, our method avoids complex homomorphic operation, thereby mitigating the computational overhead of HE. In this manner, the data remains protected from all participants in the learning process. Our findings demonstrate that, for an equivalent level of privacy, our method introduces less noise compared to the local DP method, resulting in increased accuracy after aggregation. However, the privacy amplification requires a substantial number of clients, which make our approach more suitable for cross-device Federated learning. © IFIP International Federation for Information Processing 2024
Development of a TRIZ-Based Design Methodology for Digital Transformation
No full textPart 2: Digital Transformation, Industry 4.0, and Predictive AnalyticsInternational audienceSince 2010s at business, digital transformation (DX) initiatives and digitalizing businesses have increased. The design of digital information systems is influenced by business and technological aspects and can have a considerable impact on the attractiveness of services. However, research on design science for DX or digitalizing business remains inadequate and not established. Meanwhile, the effectiveness of many methods established in the engineering field, including TRIZ that supports design and conceptual design, has been tested. The authors proposed “Creative and inventive Design Support System (CDSS)” in 2009 to support conceptual design in the engineering field based on TRIZ and suggested new methods “CDSSforDX” for applications in digital business and digital transformation (DX) in 2023. CDSSforDX is based on CDSS and has two additional processes, “strategic response” and “digital business strategy and DX strategy process,” as the strategy. In this paper, we present CDSSforDX methods and show the evaluation results and examples of using this methodology
Standardization of Engineering and Systemic Innovation
No full textPart 4: Customer Experience and Service Innovation with TRIZInternational audienceThis article offers a methodological contribution for the standardization of engineering and systemic innovation with a view to supporting design teams in the development of eco-innovative products. We propose a unified methodology that will allow users to self-guide to create eco-innovative product concepts through the application of C-K theory combined with the TRIZ method while integrating the principles of the ASIT method. Firstly, understanding the links connecting the different sources of knowledge linked to the birth of eco-innovative concepts makes it possible to design one or more generic solutions. A simplified and adapted application of TRIZ will lead us to an eco-innovative solution. Thus, to refine the results, an AI algorithm will generate new solutions from the data and conditions recorded within the artificial intelligence
Topological Optimization of a Car Module with TRIZ and Machine Learning
No full textPart 1: Sustainable and Industrial Design with TRIZInternational audienceThis study explores a methodology for the topological optimization of car modules by integrating TRIZ (Theory of Inventive Problem Solving) and machine learning techniques. Initially, TRIZ principles guide the qualitative optimization phase, establishing proper design directions aimed at weight reduction and durability enhancement. Following this, machine learning tools, including ARRK’s proprietary algorithms, are applied for precise parametric optimization, ensuring alignment with performance criteria. The findings demonstrate the efficacy of this integrated approach, significantly improving car module design by refining geometrical proportions and achieving dual objectives: weight reduction and enhanced strength. While the study highlights the potential of combining TRIZ and machine learning, it acknowledges limitations due to the use of freely available 3D models and the proprietary nature of certain algorithms. Nonetheless, this research provides a comprehensive framework for automotive engineers and designers, setting a new benchmark for incorporating qualitative insights into the quantitative optimization of complex systems
Integrated Dynamic Flexible Job Shop and AIV Scheduling: Deep Reinforcement Learning Approach Considering AIV Charging and Capacity Constraints
No full textInternational audienceScheduling Automated Intelligent Vehicles (AIV) in Dynamic Flexible Job Shops is a challenging problem due to its high level of stochasticity and dynamic nature. Various heuristic and exact methods have proven effective when the complexity of the problem is relatively low. Moreover, in the past decade, machine learning algorithms, particularly reinforcement learning, have been applied to sophisticated scheduling tasks and demonstrated their efficiency in solving such problems. In this study, a deep reinforcement learning approach is proposed to address the integrated Dynamic Flexible Job Shop and AIV scheduling. Aimed at optimizing two objectives: total lateness of jobs and total energy consumption of AIVs. The study takes into account the limitations of AIV transporters, such as charging consumption and loading capacity. To validate the proposed method, a case study of a flexible job shop is designed, and our approach is compared with a combination of existing heuristics
Towards a Sustainable Digitalization Roadmap for SMEs
No full textPart 1: Lean Thinking Models for Operational Excellence and Sustainability in the Industry 4.0 EraInternational audienceDigitalization has become a megatrend in the modern manufacturing industry. While the digitalization of production and supply chain operations arises as a challenge even for large companies, for small- and medium-sized enterprises (SMEs) the task is downright daunting. Many such firms turn to consulting firms, or academic/research institutions in the hope of finding a roadmap to the promised land. However, a review of the academic and practitioner literature leaves much to the imagination. In this paper, we present a summary of the key existing works on digitalization roadmaps, and through use of a single case study, present a more holistic, descriptive roadmap that can be used by SMEs to guide their sustainable digitalization efforts more systematically
Enhancing Labor Flexibility in Workload Control: The Development and Application of a Framework
No full textPart 1: Lean Thinking Models for Operational Excellence and Sustainability in the Industry 4.0 EraInternational audienceThis article delves into the integration of labor flexibility (LF) within Workload Control (WLC) in Make-to-Order (MTO) production settings. In a domain where existing literature offers limited guidance on data collection for optimizing LF, our study introduces the 'FlexiFlow' framework. This practical tool bridges this gap by enhancing operational efficiency and improving labor resource management and data acquisition in high-variety, low-volume MTO environments. We explore the interplay between WLC and LF through a systematic and narrative literature review. We explain effective data collection strategies, encompassing manual and digital methods, including Manufacturing Execution Systems (MES). The FlexiFlow framework, articulated through four detailed tables, equips companies with the tools to manage LF effectively, offering practical implications for practitioners. This framework extends theoretical understanding and offers actionable insights, significantly enhancing operational adaptability and efficiency. FlexiFlow improved production efficiency and responsiveness by reducing lead times and improving labor resource allocation
Management of Measuring Equipment for Quality Assurance in Manufacturing Processes: A Decision-Making Support System
No full textPart 1: Modelling Supply Chain and Production SystemsInternational audienceThis paper aims to present a decision-making support system for the efficient management of measuring equipment in Manufacturing plants, providing support for condition analysis and calibration interval definition, a missing functionality in current computerized systems. For the same equipment, existing methods for defining the calibration/verification interval give very different results. The proposed solution for management of measuring equipment includes a developed approach for the selection of the method that best suits the equipment, considering its characteristics and condition of use. Since unpredictable situations, such as falls, shocks or misuse can change the behavior of measuring equipment, an additional approach for unpredictable behavior identification is included, using EWMA and I-MR charts. The approaches were developed and tested with data of measuring equipment of a cork company. This solution brings cost savings to companies since, by helping to adjust calibration or verification intervals, allows avoiding the use of inapt equipment for quality control of manufactured products
Operations Management of Additive Manufacturing
No full textPart 5: Additive ManufacturingInternational audienceThis article reviews the growing literature on additive manufacturing (AM) operations management and sheds light on the emerging research areas in this field. As the AM use cases of final parts rapidly expand, it is essential to focus on the operations management of this technology and determine the primary current and future research streams. A literature study method is utilized to select, review, and categorize articles in the field of AM. The 108 articles selected after the initial evaluation were carefully examined and categorized. The selected papers evaluate AM from an operations management perspective. This article categorizes the body of knowledge studying the application and operations management of additive manufacturing into three categories: studies concerned with the industry's current state, forward-looking studies with a conceptual approach, and forward-looking papers with empirical grounding. Different AM processes studied are also considered. Our categorization showed that the latter category is still under-researched and presents an opportunity for future investigations. Moreover, six emerging streams of research in the third category were recognized. In addition to pointing out the areas of research that require more attention, this article aims to assist the researchers in better positioning their research
Assembly Line Design for Industrialized Electrolyser Production
No full textPart 1: Modelling Supply Chain and Production SystemsInternational audienceThe electrolyser industry is under rapid development, where manufacturing companies seek opportunities to develop industrialized solutions for electrolyser production. However, electrolyser production implies major logistical challenges related to for instance product characteristics (expensive and fragile components, rare materials), the stack assembly process (precise alignment of cells and pressure), the supply chain (immature, limited supplier base, quality issues, delivery constraints), and the production equipment (immature). Designing a high efficiency electrolyser assembly line is thus a challenging task. This paper presents results of a study with the aim to assess design alternatives for electrolyser stack assembly with focus on logistics and flow efficiency. Design elements such as the number and locations of buffers and quality inspection are considered. A discrete-event simulation model is developed and tested based on scenarios reflecting situations related to for instance poor quality yield of incoming components and equipment downtime. Data from an empirical case involving the design of a new assembly line, are used to set up the model. Based on the findings from the simulations, recommendations for the design of industrialized electrolyser assembly lines are formulated. This study contributes with insight to critical aspects and guidance to design decisions for efficient electrolyser assembly lines. Moreover, it shows how simulation can support assessment of different design alternatives in the development of efficient stack assembly lines