Big Data Analytics and Intelligent Operations & Maintenance
课程组介绍:
主讲人：Eugene Chen，曼尼托巴大学助理教授、博士生导师。2020年获得加拿大阿尔伯塔大学QS-92机械工程专业博士学位，师从加拿大工程院院士左明健教授，而后在该校任职博士后至2021年；2021年-2024年期间任职同济大学载运工具运用工程专业特聘研究员、博导；2024年至今入职加拿大曼尼托巴大学。研究方向包括：状态监测、智能运维、物理信息神经网络等。已先后主持了加拿大自然科学与工程研究基金、中国国家自然科学基金、上海市人才项目、国重实验室开放课题、中车中船等企业委托项目；参与了国家自然科学基金面上项目，加拿大的自然科学基金Discovery Grant、重点研发、Mitacs校企合作、Mitacs国际交流项目。在IEEE Trans Ind Informat、Mech Syst Signal Pr、J Sound Vib等国际高水平期刊发表学术论文40余篇，其中一作/通讯38篇，担任IEEE Ins Mea、IEEE Sens J、Int J Progn Health M、Mechat Sys Ctrl J期刊编委，Mech Syst Signal Pr、Reliab Eng Syst Safe等7个国际高水平期刊的客座编辑，JDMD、城市轨道交通研究期刊青年编委，获授权中国发明专利12项。另外，陈博士还获得过阿尔伯塔未来科技创新奖、加拿大旋转机械年会最佳论文奖、国际预测与健康管理数据分析竞赛奖等科研奖励。
Personal Profile:
Eugene Chen, Assistant Professor and Ph.D. Supervisor at the University of Manitoba. He earned his Ph.D. in Mechanical Engineering (QS-92) from the University of Alberta, Canada, in 2020 under the supervision of Professor Ming J. Zuo, a Fellow of the Canadian Academy of Engineering. He then served as a postdoctoral researcher at the same university until 2021. From 2021 to 2024, he was appointed as a Distinguished Researcher and Ph.D. Supervisor in the field of Vehicle Operation Engineering at Tongji University. Since 2024, he has been working at the University of Manitoba, Canada. His research focuses on condition monitoring, intelligent operation and maintenance, physics-informed neural networks, and related areas.
He has led multiple research projects, including grants from the Natural Sciences and Engineering Research Council of Canada (NSERC), the National Natural Science Foundation of China, Shanghai Talent Programs, open projects from national key laboratories, and industry-sponsored projects from companies such as CRRC and CSSC. He has also participated in projects such as the NSFC General Program, Canada’s NSERC Discovery Grant, key R&D initiatives, Mitacs industry-academic collaborations, and Mitacs international exchange programs.
With over 40 publications in top-tier international journals such as IEEE Transactions on Industrial Informatics, Mechanical Systems and Signal Processing, and Journal of Sound and Vibration, he has served as the first or corresponding author for 38 of them. He is an editorial board member for journals including IEEE Instrumentation & Measurement Magazine, IEEE Sensors Journal, International Journal of Prognostics and Health Management, and Mechatronic Systems and Control Journal, as well as a guest editor for seven high-impact journals such as Mechanical Systems and Signal Processing and Reliability Engineering & System Safety. Additionally, he is a young editorial board member for JDMD and Urban Rail Transit Research and holds 12 authorized Chinese invention patents.
Dr. Chen has received several academic awards, including the Alberta Future Innovates Award, the Best Paper Award at the Canadian Rotating Machinery Conference, and the PHM Data Analysis Competition Award.
课程负责人：冯珂，西安交通大学教授、博士生导师，国家级青年人才、玛丽居里学者、全球前2%顶尖科学家，本硕毕业于电子科技大学，博士毕业于新南威尔士大学。曾在英属哥伦比亚大学、新加坡国立大学、帝国理工学院等知名学府任职。研究方向涵盖数字孪生、信号处理、故障诊断、疲劳磨损分析等领域。2023年荣获皇家物理协会会刊评选的“新锐科学家”称号。现担任《IEEE Transactions on Industrial Informatics》、《Information Fusion》、《Structural Health Monitoring》等多个国际期刊的副编辑及编委。研究成果发表在《IEEE Transactions on Fuzzy Systems》、《Mechanical Systems and Signal Processing》等重要学术期刊上。主持多项国际合作项目，包括欧盟“地平线”项目、英国研究与创新署项目等，及国家自然科学基金优青项目（海外）和国家重点研发计划课题等，曾获得“中国航空学会科技奖二等奖” 和“振动工程学会科学技术奖二等奖”。
Profile:
Ke Feng, Professor at Xi'an Jiaotong University, is a Marie Curie Fellow and ranked among the “Stanford/Elsevier Top 2% Scientists”. He earned his bachelor's and master's degrees from the University of Electronic Science and Technology of China and his Ph.D. from the University of New South Wales. He has held positions at renowned institutions such as the University of British Columbia, the National University of Singapore, and Imperial College London. His research areas include digital twins, signal processing, fault diagnosis, fatigue, and wear analysis, among others. In 2023, he was awarded the title of “Emerging Leader” by the Royal Physical Society Journal. He currently serves as an Associate Editor and Editorial Board Member for several international journals, including IEEE Transactions on Industrial Informatics, Information Fusion, and Structural Health Monitoring. His research findings have been published in prestigious academic journals such as IEEE Transactions on Fuzzy Systems & Mechanical Systems and Signal Processing. He has led numerous international collaborative projects, including the Horizon Europe, UKRI projects, the National Natural Science Foundation Excellent Young Scientists Fund (Overseas), and the key projects under the National Key Research and Development Program of China. He has also received the “Second Prize of the China Aviation Science and Technology Award” and the “Second Prize of the Vibration Engineering Society Science and Technology Award”.
课程简介：
第一次课程：
时间: 2025年7月21日 14:00-15:30 北京时间
线上：Zoom会议链接
https://ubc.zoom.us/j/66228723478?pwd=DdGypfGBgyl0JIUrhdK9zOSfVIRH9U.1
会议号: 662 2872 3478
密码: 593573
第一次：课程题目
物理知识驱动机器学习及其装备状态监测方面的应用
内容简介
深度学习相比经典机器学习具有更强的学习能力，尤其是在数据量增大的情况下。然而，纯数据驱动的机器学习模型有时可能无法遵循物理定律，且缺乏决策透明性和可解释性。
物理知识驱动的机器学习（Physics-Informed Machine Learning，PIML）通过将物理知识和约束集成到机器学习过程中，解决了这些问题。该方法结合了领域特定的物理洞察和先进的机器学习技术，提升了模型的性能，并确保预测结果与已知的物理原理保持一致。在本次报告中，我们将探讨物理知识驱动的机器学习领域的若干贡献，包括：
· 物理知识驱动的超参数选择：通过引入物理约束来增强模型参数调优的方法。
· 显式速度集成的循环神经网络（RNN）：通过将转速决定振动特性这一物理定律融入到RNN结构设计的方式，改善RNN的准确性和稳定性。
· 物理知识驱动的残差建模：通过纠正机器学习预测与物理模型之间的偏差，确保更可靠的回归输出。
我们还将讨论这些技术在旋转机械和铁路车辆中的应用，展示如何通过融合物理知识来增强这些领域的预测能力。
First Lecture
Time: July 21, 2025, 14:00–15:30 (Beijing Time)
Online: Zoom Meeting
Link: https://ubc.zoom.us/j/66228723478?pwd=DdGypfGBgyl0JIUrhdK9zOSfVIRH9U.1
Meeting ID: 662 2872 3478
Password: 593573
Lecture Title:
Physics-Informed Machine Learning and Its Applications in Equipment Condition Monitoring
Abstract:
Deep learning exhibits stronger learning capabilities than classical machine learning, particularly with large datasets. However, purely data-driven machine learning models may sometimes violate physical laws and lack transparency and interpretability in decision-making.
Physics-Informed Machine Learning (PIML) addresses these challenges by integrating physical knowledge and constraints into the machine learning process. This approach combines domain-specific physical insights with advanced machine learning techniques to enhance model performance and ensure predictions align with established physical principles.
In this lecture, we will explore several contributions in the field of physics-informed machine learning, including:
· Physics-Informed Hyperparameter Selection: A method to enhance model parameter tuning by incorporating physical constraints.
· Explicit Speed-Integrated Recurrent Neural Networks (RNNs): An RNN architecture that incorporates the physical law linking rotational speed to vibration characteristics, improving accuracy and stability.
· Physics-Informed Residual Modeling: A technique to correct deviations between machine learning predictions and physical models, ensuring more reliable regression outputs.
We will also discuss applications of these techniques in rotating machinery and railway vehicles, demonstrating how integrating physical knowledge enhances predictive capabilities in these fields.
第二次课程：
时间: 2025年7月22日 10:00-11:30北京时间
线上：Zoom会议链接
https://ubc.zoom.us/j/61668812247?pwd=0KSCBtkBgTcK6vYHRQjh7TKz8ZYuS8.1
会议号: 616 6881 2247
密码: 865917
第二次：课程题目
旋转机械状态监测与故障诊断的时序表征模型方法
内容简介
旋转机械如齿轮、轴承、转子是高端装备的基础部件，其状态监测与故障诊断关乎整体装备的安全可靠高效运行。从状态监测信号的时序依赖关系角度出发，介绍面向旋转机械状态监测与故障诊断的变参数时序表征方法研究进展及未来研究方向。首先介绍面向旋转机械动力学系统辨识的时序表征建模的基本概念、基于时序表征模型参数或残差的状态监测与故障诊断框架；其次提出旋转机械变工况下非平稳状态监测信号的稀疏线性变参数时序表征模型，以及随机变工况下基于稀疏线性变参数时序表征模型的故障诊断机制；然后就非线性时序表征，提出改进变索引系数自回归模型；最后展望时序表征方法的未来研究方向。
Second Lecture:
Time: July 22, 2025, 10:00–11:30 (Beijing Time)
Online: Zoom Meeting
Link: https://ubc.zoom.us/j/61668812247?pwd=0KSCBtkBgTcK6vYHRQjh7TKz8ZYuS8.1
Meeting ID: 616 6881 2247
Password: 865917
Lecture Title:
Time-Dependent Representation Modeling for Condition Monitoring and Fault Diagnosis of Rotating Machinery
Abstract:
Rotating machinery components such as gears, bearings, and rotors are fundamental to high-end equipment, where their condition monitoring and fault diagnosis are critical for ensuring safe, reliable, and efficient operation. This lecture explores advances and future directions in time-dependent representation modeling for condition monitoring and fault diagnosis of rotating machinery, with a focus on the temporal dependencies in monitoring signals.
Key topics include:
1. Fundamentals of Time-Dependent Representation Modeling
o Basic concepts for system identification in rotating machinery dynamics.
o A framework for condition monitoring and fault diagnosis based on model parameters or residuals.
2. Sparse Linear Time-Varying Parameter Models for Nonstationary Signals
o A proposed model for handling nonstationary condition monitoring signals under variable operating conditions.
o A fault diagnosis mechanism based on sparse linear time-varying parameter models under stochastic operational variations.
3. Nonlinear Time-Dependent Representation: Improved Variable Index Coefficient Autoregressive Model
o An enhanced approach for nonlinear time-series representation to improve diagnostic accuracy.
4. Future Research Directions in Time-Dependent Representation Methods
o Emerging trends and potential advancements in the field.
This lecture will provide insights into cutting-edge methodologies for improving the reliability and efficiency of rotating machinery through advanced signal processing and modeling techniques.
Big Data Analytics and Intelligent Operations & Maintenance
Group
