Machine Learning Approaches for Diagnostics and Prognostics of Industrial Systems Using Open Source Data from PHM Data Challenges: A Review
Hanqi Su, Jay Lee
TL;DR
This review addresses the lack of unified guidelines for applying machine learning to Prognostics and Health Management (PHM) by analyzing open-source industrial datasets from PHM Data Challenge competitions conducted between 2018 and 2023. It systematically catalogs problems, datasets, tasks, and ML methods across nine competitions, and proposes a unified ML framework for PHM that encompasses data collection, processing, visualization, modeling (conventional ML and DL), and interpretability. The paper highlights recurring data- and model-related challenges, such as missing data, data imbalance, domain shift, model selection, interpretability, and robustness, and discusses strategies including transfer learning, domain adaptation, data augmentation, and XAI tools. It also identifies limitations of current competitions, notably the underutilization of multi-modal data and systematic PHM system design, and outlines future directions, including open multi-modal datasets, MMML, improved interpretability, advanced transfer/adaptation techniques, and the use of large knowledge models. Overall, the work provides practical guidance for researchers and industry to advance ML-driven PHM in real-world settings, bridging academia and practice.
Abstract
In the field of Prognostics and Health Management (PHM), recent years have witnessed a significant surge in the application of machine learning (ML). Despite this growth, the field grapples with a lack of unified guidelines and systematic approaches for effectively implementing these ML techniques and comprehensive analysis regarding industrial open-source data across varied scenarios. To address these gaps, this paper provides a comprehensive review of ML approaches for diagnostics and prognostics of industrial systems using open-source datasets from PHM Data Challenge Competitions held between 2018 and 2023 by PHM Society and IEEE Reliability Society and summarizes a unified ML framework. This review systematically categorizes and scrutinizes the problems, challenges, methodologies, and advancements demonstrated in these competitions, highlighting the evolving role of both conventional machine learning and deep learning in tackling complex industrial tasks related to detection, diagnosis, assessment, and prognosis. Moreover, this paper delves into the common challenges in PHM data challenge competitions by emphasizing data-related and model-related issues and evaluating the limitations of these competitions. The potential solutions to address these challenges are also summarized. Finally, we identify key themes and potential directions for future research, providing opportunities and prospects for next-generation ML-PHM development in PHM domain.