Reproduce “Explainable Predictive Maintenance: A Survey of Current Methods, Challenges and Opportunities” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Logan Cummins, Alex Sommers, Somayeh Bakhtiari Ramezani, Sudip Mittal, Joseph Jabour, Maria Seale, Shahram Rahimi
A systematic review of 102 articles identifies and categorizes the current landscape of Explainable Predictive Maintenance (XPM), detailing methods, challenges, and opportunities to foster trust and understanding in AI-driven maintenance for critical industrial systems. It reveals a clear increase in XPM research since 2017, with a slight preference for post-hoc Explainable AI over intrinsically interpretable machine learning.
Key insights
- The field of Explainable Predictive Maintenance (XPM) is experiencing rapid growth, with a clear increasing trajectory in publications, especially since 2017, driven by the deployment of complex AI in critical industrial settings.
- Post-hoc Explainable AI (XAI) methods are slightly more popular than intrinsically Interpretable Machine Learning (iML) methods in XPM, reflecting a tendency to apply explanations to high-performing black-box models.
- Anomaly Detection is the most frequent PdM task utilizing XAI and iML, with XAI often preferred for prognosis and iML showing stronger interest in fault diagnosis.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.