Special Session #01 AI-Enabled Prognostics and Health Management for Renewable Energy Systems
Special Session #04 Multivariate signal analysis and multimodal data fusion in PHM applications
Special Session #05 Model-Data Interaction for diagnostics and prognostics of Mechanical Systems
AI-Enabled Prognostics and Health Management for Renewable Energy Systems
Session Organizers:
Dandan Peng, The Hong Kong Polytechnic University
Email: dandanpeng2@gmail.com
Jinpeng Tian, The Hong Kong Polytechnic University
Email: jinpeng.tian@polyu.edu.hk
Zhuyun Chen, Guangdong University of Technology
Email: mezychen@gdut.edu.cn
Download: Special Session #1.pdf
Renewable energy systems, such as wind turbines, photovoltaic (PV) systems, and energy storage systems, are critical components of the global energy transition. However, these systems often operate in complex and dynamic environments, making them susceptible to performance degradation and failures due to factors such as weather conditions, load fluctuations, and aging equipment. Traditional maintenance strategies, which are typically reactive or schedule-based, are inefficient and costly. The integration of Artificial Intelligence (AI) and Prognostics and Health Management (PHM) technologies offers a transformative approach to monitor, predict, and manage the health of renewable energy systems. By leveraging AI-driven data analytics, multi-sensor fusion, and predictive modeling, PHM enables real-time fault diagnosis, remaining useful life (RUL) prediction, and optimized maintenance strategies, thereby improving system reliability and reducing operational costs. This special session aims to bring together researchers, engineers, and industry experts to present and discuss the latest advancements in AI-enabled PHM for renewable energy systems.
The topics of interest include, but are not limited to:
• Multi-sensor data fusion
• Machine learning and deep learning algorithms for anomaly detection, fault classification, and failure prediction
• Transfer learning and domain adaptation for fault diagnosis and RUL prediction
• Digital twin technology for virtual representation and real-time monitoring
• Explainable AI (XAI) for interpretable fault diagnosis and decision-making
• Physics-informed neural networks for degradation modeling and RUL prediction
• Uncertainty quantification and risk assessment
• Fleet-level condition monitoring
• Any other related topics
Graph Representation Learning for Prognostics and Health Management in Industrial Systems: Towards Intelligent Maintenance
Session Organizers:
Jie Liu, Huazhong University of Science and Technology
Email: jie_liu@hust.edu.cn
Xingxing Jiang, Soochow University
Email: jiangxx@suda.edu.cn
Yanglong Lu, The Hong Kong University of Science and Technology
Email: maeylu@ust.hk
Zhongxu-Hu, Huazhong University of Science and Technology
Email: zhongxu_hu@hust.edu.cn
Download: Special Session #2.pdf
Prognostics and health management (PHM) has attracted significant attention within the industrial community due to its potential to enhance equipment reliability and reduce operational costs. However, several challenges remain in effectively managing existing knowledge to identify the cause-and-effect relationships between failures, particularly in establishing the correlation between identified failures and their root causes. Graph representation learning (GRL), with its focus on relationship modeling, presents a promising approach to address these challenges. By capturing and leveraging the dependencies between system components or equipment, GRL demonstrates remarkable potential in advancing PHM tasks and opens new avenues for model interpretability. Despite these advantages, the implementation of GRL varies significantly across different PHM tasks, making it challenging to develop standardized graph construction strategies for diverse scenarios. Given the current lack of unified frameworks and guidelines for applying GRL to specific PHM tasks, further investigations are needed to bridge this gap. This special session aims to foster the development of GRL for PHM applications. Therefore, we strongly encourage researchers to submit original research and review articles that contribute to this critical area of study.
Potential topics include but are not limited to the following:
Mechanism Knowledge Embedded Intelligent Mechanical Fault Diagnosis and Remaining Useful Life Prediction
Session Organizers:
Fei Jiang, Dongguan University of Technology
Email: jiangfei@dgut.edu.cn
Gang Chen, South China University of Technology
Email: gangchen@scut.edu.cn
Jipu Li, The Hong Kong Polytechnic University
Email: jipu1994.li@polyu.edu.hk
Zuozhou Pan, China Jiliang University
Email: panzz@cjlu.edu.cn
Download: Special Session #3.pdf
Modern mechanical systems often operate under complex conditions involving variable speeds, dynamic loads, and high noise levels, which obscure critical fault-related features in raw vibration signals. While data-driven approaches like deep learning have shown promise in fault diagnosis (FD) and remaining useful life (RUL) prediction, their performance is constrained by insufficient physical interpretability, poor generalization under unknown conditions, and limited integration with domain-specific mechanism knowledge. To address these challenges, embedding physical principles of mechanical degradation (e.g., fatigue dynamics, tribological interactions, and failure propagation mechanisms) into intelligent models has emerged as a transformative strategy. This Special Session focuses on methodologies that synergize mechanism-driven knowledge with advanced data analytics to enhance model robustness, interpretability, and cross-domain adaptability. Topics include physics-guided signal processing, mechanism-embedded hybrid deep learning architectures, and knowledge-informed degradation modeling, aiming to bridge the gap between theoretical mechanics and intelligent prognostics for real-world industrial applications.
The topics of interest include, but are not limited to:
• Physics-guided signal processing for fault feature extraction
• Mechanism-driven degradation modeling and RUL prediction
• Interpretable hybrid models combining physics and deep learning
• Embedding tribological/fatigue mechanisms into data-driven frameworks
• Knowledge-informed sensor fusion for multi-source data analytics
• Mechanism-aware domain adaptation under variable operating conditions
• Physics-constrained neural networks for FD and RUL prediction
• Real-time monitoring systems integrating mechanism knowledge
• Case studies on rotating machinery, bearings, gears, and aerospace systems
• Benchmark datasets and evaluation metrics for physics-aware models
Multivariate signal analysis and multimodal data fusion in PHM applications
Session Organizers:
Prof. Zong Meng, Yanshan University
Email: mzysu@ysu.edu.cn
Prof. Yuejian Chen, University of Manitoba
Email: Yuejian.Chen@umanitoba.ca
Associate Prof. Rui Yuan, Wuhan University of Science and Technology
Email: yuanrui@wust.edu.cn
Postdoctoral Associate. Hewenxuan Li, Cornell University
Email: hewenxuan.li@cornell.edu
Download: Special Session #4.pdf
Prognostic and Health Management (PHM) undergoes paradigm-shifting transitions from reactive maintenance to proactive and self-aware intelligence in industrial systems. The indispensable thrust of fusing multivariate signal analysis and multimodal data fusion will tackle challenges of modern equipment PHM. Multivariate signal analysis focuses on analyzing high-dimensional, multi-channel, and multi-sensor data. This analysis allows for the extraction of critical features that are sensitive to degradation, even in the presence of noise, interference, and changing operating conditions. By processing these signals, advanced techniques can isolate potential faults early on, providing a precise model of how the system’s condition is degrading over time. This early detection is essential for preventing unexpected failures and extending the lifespan of the equipment. Meanwhile, multimodal data fusion also plays a vital role in combining information from various types of sensors to form a comprehensive understanding of the system’s health. Multi-modal sensory inputs provide diverse insights into system health but simultaneously subjected to various uncertainties. AI-driven data fusion techniques can enhance health indicators while quantifying uncertainties that reveal equipment’s state. These capabilities underpin the transformative PHM transitions: from sensor-based diagnostics to AI-driven prognostics, scheduled to condition-based maintenance, isolated to holistic lifecycle management. This session highlights innovations bridging sensing, analytics, and action. Applications span aerospace, energy, and smart manufacturing.
The topics of interest include, but are not limited to:
Session Organizers:
Prof. Qing Ni, Northwestern Polytechnical University
Email: qing.ni@outlook.com.au
Associate Prof. Ying Zhang, University of Science and Technology Beijing
Email: ying.zhang@ustb.edu.cn
Associate Prof. Chenyang Ma, Xi’an University of Posts & Telecommunications
Email: machenyang@xupt.edu.cn
Download: Special Session #5.pdf
Mechanical systems are the backbone of critical industries, including manufacturing, energy, transportation, and aerospace. The failure of key components, such as bearings, gears, and turbines, in mechanical systems can lead to catastrophic consequences, including substantial economic losses, safety risks, and operational downtime. Traditional fault diagnosis and remaining useful life (RUL) prediction methods typically rely on either physics-based models or purely data-driven approaches, both of which with inherent limitations. Physics-based models offer interpretability and domain knowledge but may struggle to generalize across diverse and dynamic operating conditions. On the other hand, data-driven methods often require extensive labeled datasets and may lack physical consistency, limiting their reliability in real-world applications.
The integration of model-based and data-driven approaches, often referred to as model-data interaction, has emerged as a promising paradigm to overcome these limitations. By combining the interpretability and domain knowledge of physics-based models with the adaptability power of data-driven techniques, this fusion enables more accurate, robust, and interpretable fault diagnosis and RUL prediction. This session aims to explore the latest advancements, challenges, and applications of model-data interactive fault diagnosis and RUL prediction methods for key components of mechanical systems in complex and dynamic operating environments.
The topics of interest include, but are not limited to:
Prof. Hao Zhang, Hebei University of Technology
Email: zhanghao@hebut.edu.cn
Prof. Zhaozong Meng, Hebei University of Technology
Download: Special Session #6.pdf
Rotating machinery is one of the most common classes of mechanical equipment and plays an important role in industrial applications. It generally operates under tough working environment and is therefore subject to failures, which may cause machinery to break down and decrease machinery service performance such as manufacturing quality, operation safety, etc. It is therefore necessary to carry out an efficient condition monitoring for rotating machinery to increase reliability against possible faults. Traditional monitoring techniques include lubricant oil analysis, vibration and noise analysis, thermal analysis, etc. Lubricant oil analysis is generally offline and needs special instruments. Vibration and noise analysis are susceptible to the background noise. Thermal analysis is often not sensitive to the fault until reaching a serious stage.
Adopting novel smart sensing technology to acquire signals that include rich information regarding the operating condition of rotating machinery will be meaningful from the diagnostic point of view. In addition, AI-based condition monitoring approaches have become noteworthy for interpretating the acquired signals and obtaining fault signatures as early as possible. This session aims to provide a common platform for professionals, engineers, practitioners and researchers to explore their latest achievements in the field of smarting sensing technology and AI-based condition monitoring for rotating machinery.
The topics of interest include, but are not limited to:
Prof. Zhongliang Xie, Northwestern Polytechnical University
Email: zlxie@nwpu.edu.cn
Prof. Wenjun Gao, Northwestern Polytechnical University
Email: gaowenjun@nwpu.edu.cn
Associate Prof. Huihui Feng, Hohai University
Email: fenghh@hhu.edu.cn
Download: Special Session #7.pdf
The mechanical transmission system is the core power heart of large critical equipment, including large ships, aircraft engines, turbines, and so on. As the key supporting component of mechanical systems, bearings are the "joints" of the entire mechanical transmission system. Its lubrication performance, load-bearing capacity, fault diagnosis and prediction will directly affect the safe, stable and reliable operation of the entire mechanical transmission system.
However, the traditional bearings (whether plain journal bearings or rolling element bearings) have limitations in terms of load-bearing capacity, working range, vibration and noise reduction performances. Therefore, there is an urgent need to develop and propose new configurations of bearings——Intelligent bearings. In addition, the lubrication theory, fault prediction methods, and model theoretical system of intelligent bearings also urgently need to be expanded. Based on this, the topic will explore new theories of bearing lubrication, accurate prediction models for load-bearing capacity, fault diagnosis methods and prediction models, etc.
The topics of interest include, but are not limited to:
https://icsmd2025.aconf.org/
icsmd2025@163.com