Session Aims & Scope
With the powerful capabilities, digital twin is enabling unprecedented innovation in various industries. The digital twin applications become more and more extensive, covering almost every industry. In the future, the development and application of digital twin will pay more attention to interdisciplinary integration and innovation. Through the continuous introduction of new technologies and methods, digital twin enables the complete optimization and control of industrial production processes, to provide strong support for the sustainable development of industrial enterprises. This session aims to explore the transformative potential of digital twin in the interdisciplinary integration and innovation. It will focus on innovative applications of digital twin in various industries, emphasizing how interdisciplinary technologies can help digital twin continuous development. The session seeks to build bridge for different researchers, practitioners of digital twin from various disciplinary, providing a platform for stakeholders to discuss advancements, share insights, and foster collaborations that propel the industry forward.
The topics of session is not limited. Participants will include researchers, technology developers, industry practitioners, and policymakers, who will delve into technological innovations and strategic frameworks essential for advancing digital twin in multidisciplinary innovation.
Session Chair(s)
Chair
Fei Tao
Professor, Beihang University (China)
Co-Chair
Nabil ANWER
Paris-Saclay University (France)
CIRP Fellow, AET Fellow
Co-Chair
Qinglin Qi
Associate Professor , Beihang University (China)
kylin3366@buaa.edu.cn
Co-Chair
Gang Hou
Professor , China-Japan Friendship Hospital (China)
hougangcmu@163.com
Session Presentation
1.
Yang Shi
Professor
IEEE, ASME, EIC, CSME Fellow, University of Victoria (Canada)
Title: Advanced Robust Model Predictive Control (MPC) Framework for Autonomous Intelligent Mechatronic Systems
Abstract
Networked and distributed control for mechatronic systems have received great attention in the control community due to its wide application areas. Network-induced limitations may be caused by the presence of a communication channel, or because of the efficient assignment of power and other limited resources. Intelligent mechatronic systems represent a large class of smart systems that encompass computational (i.e., hardware and software) and physical components, seamlessly integrated and closely interacting to autonomously sense and manipulate the changing state of the physical system. These systems involve a high degree of complexity at numerous spatial and temporal scales and highly networked communications integrating computational and physical components. Model predictive control (MPC) is a promising paradigm for high-performance and cost-effective control of networked and distributed mechatronic systems. This talk will firstly summarize the major application requirements and challenges to innovate in designing, implementing, deploying and operating intelligent mechatronic systems. Further, the robust MPC and distributed MPC design methods will be presented. Finally, the application of MPC algorithms to a variety of autonomous intelligent mechatronic systems will be illustrated.
2.
Jun Qian
Research manager, Department of Mechanical Engineering, KU Leuven (Belgium)
Core lab manager, Manufacturing and Automation corelab, Flanders Make (Belgium)
Title: Digital twin of manufacturing processes for enhanced process stability and efficiency
Abstract
Process efficiency, quality and costs are the three key elements in manufacturing. Digital twins of manufacturing processes, based on the theoretical process models and enriched with in-process measurement data of critical parameters, have demonstrated promising effects on improving the process stability, its efficiency and quality prediction. Two case studies, namely blind hole drilling by electrical discharge machining (EDM) and mechanical milling, are explained in this presentation. In the EDM drilling process, machine learning is applied to update the digital twin of the process.
3.
Min Xia
Associate Professor and the Director
The University of Western Ontario (Canada)
Title: Trustworthy AI-based Solutions for Industrial Machine Condition Monitoring
Abstract
Deep neural networks have been widely investigated in data-driven machine condition monitoring due to their superior capabilities in classification and regression. However, the “black box” nature of deep learning has limited the practical application of these methods in real industrial settings. Building reliable and trustworthy approaches is both an urgent and demanding task in academia and industry. This presentation will illustrate trustworthy AI-based methods, focusing on interpretable model learning and uncertainty estimation, which can significantly enhance the reliability of AI-based solutions for industrial machine condition monitoring. Through real-world case studies, the research and future application scenarios of these methods will be explored.
4.
Qiuhua Huang
Associate Professor
Colorado School of Mines (U.S.)
Title: Convergence of AI, Physics, and Computing for Intelligent Emergency Control in Large-Scale Power Systems
Abstract
This presentation explores a cutting-edge convergence framework for intelligent emergency control in large-scale power systems, which leverages the integration of machine learning, physics, advanced computing, and grid control. Power systems are operating closer to their limits with increasing uncertainties, posing new challenges to stability and security, especially during extreme events like hurricanes and heat waves. The framework is exemplified through a deep reinforcement learning (DRL)-based approach, which has been applied to a synthetic Texas power system of over 3,000 buses and tested with more than 56,000 scenarios. The proposed hybrid system demonstrated a 26% reduction in load shedding, outperforming traditional rule-based controls in nearly 100% of cases, highlighting the effectiveness of AI for grid control. This presentation will offer attendees insights into the future of AI-driven digital twins for grid operations, with an emphasis on scalability, adaptability, and security in real-time control environments. The convergence framework presented can serve as a model for developing resilient, intelligent grid systems of the future.
5.
Qi Liu
Lecturer
University of Bath (UK)
Title: Predictive digital twin-driven dynamic error control for ultra-precision machining: A Case study in slow-tool-servo ultra-precision diamond turning
Abstract
A predictive digital twin (DT)-driven dynamic error control approach is presented for accuracy control in ultra-precision machining. Taking slow-tool-servo ultra-precision diamond turning processes as a demonstrator, an explainable artificial intelligence-enabled real-time DT of the total dynamic error (inside and outside the servo loop) was established using in-line acceleration input data near the tool. A feedforward controller was used to mitigate the total dynamic errors before they came into effect. The machining trials using this approach showed that significant improvement in machining accuracy (87%, surface form accuracy; 95%, phase accuracy with precisions of 0.06 µm and 0.05°), and efficiency (8 times the state-of-the-art) were successfully achieved.
6.
Pingchuan Ma
Postdoctor
The Imperial College London (UK)
Title: Deep Audio-Visual Speech Recognition
Abstract
Automatic Speech Recognition (ASR) is the task of recognizing speech by using audio information. Recent advances in deep learning have substantially improved speech recognition technology, leading to widespread commercial applications across various scenarios. However, even the most advanced ASR systems struggle with robustness in noisy environments. A solution to mitigate this issue is the incorporation of visual information alongside audio data, as the visual stream is not affected by background noise. This presentation will explore the integration of audio-visual speech recognition (AVSR). It will begin by discussing recent progress in the field of visual speech recognition, commonly referred to as lipreading. Then, it will highlight how cross-modal learning can further enhance lipreading performance. Finally, the presentation will introduce self-supervised learning methods for joint audio-visual representation and discuss potential trends for future developments.
7.
Gangming Zhao
PhD Candidate
The University of Hong Kong (China)
Title: Diagnose Like a Radiologist: Hybrid Neuro-Probabilistic Reasoning for Attribute-Based Medical Image Diagnosis
Abstract
Effectively modeling attributes as well as all relationships involving attributes could boost the generalization ability and verifiability of medical image diagnosis algorithms. In this topic, we introduce a hybrid neuro-probabilistic reasoning algorithm for verifiable attribute-based medical image diagnosis. There are two parallel branches in our hybrid algorithm, a Bayesian network branch performing probabilistic causal relationship reasoning and a graph convolutional network branch performing more generic relational modeling and reasoning using a feature representation. Tight coupling between these two branches is achieved via a cross-network attention mechanism and the fusion of their classification results. We have successfully applied our hybrid reasoning algorithm to two challenging medical image diagnosis tasks. On the LIDC-IDRI benchmark dataset for benign-malignant classification of pulmonary nodules in CT images, our method achieves a new state-of-the-art accuracy of 95.36% and an AUC of 96.54%. Our method also achieves a 3.24% accuracy improvement on an in-house chest X-ray image dataset for tuberculosis diagnosis.
8.
Yulong Zhao
Postgraduate student
Wuhan University of Science and Technology (China)
Title: Ensemble Learning-Based Stability Improvement Method for Feature Selection towards Performance Prediction
Abstract
The uncertainty and complexity of real data collected in the industrial production process increase the difficulty in data-based knowledge discovering. Feature selection is an important step to remove redundant and irrelevant data, and thus it is essential to construct an efficient feature selection method. In this paper, an ensemble learning-driven stable feature selection method is proposed to improve the stability and accuracy of the feature selection. Firstly, datasets of different characteristics are generated to increase the diversity of data segments for feature selection. Secondly, two criteria (stability and prediction accuracy) are adopted to evaluate the performance weight of each feature selection algorithm, to ensure that the results of high-performance selectors have high priority in the algorithm aggregation process. Thirdly, the feature subsets are weighted and filtered based on expert experience to further ensure its stability. Finally, comparative experiments are conducted to show the effectiveness of the proposed method.