Explainable Predictive Maintenance of Rotating Machines Using LIME, SHAP, PDP, ICE

• Published in: IEEE access Vol. 12; pp. 29345 - 29361
• Main Authors: Gawde, Shreyas, Patil, Shruti, Kumar, Satish, Kamat, Pooja, Kotecha, Ketan, Alfarhood, Sultan
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Artificial intelligence; Classification algorithms; Critical components; Data acquisition; Decisions; Explainable AI; Explainable artificial intelligence; Feature extraction; Fourth Industrial Revolution; ICE; Industrial applications; industrial rotating machines; Industry 4.0; LIME; PDP; Prediction algorithms; Predictive maintenance; Predictive models; Rotating machinery; Rotating machines; SHAP
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2024.3367110

Artificial Intelligence (AI) is a key component in Industry 4.0. Rotating machines are critical components in manufacturing industries. In the vast world of Industry 4.0, where an IoT network acts as a monitoring and decision-making system, predictive maintenance is quickly gaining importance. Predictive maintenance is a method that uses AI to handle potential problems before they cause breakdowns in operations, processes or systems. However, there is a significant issue with the AI models' (also known as "black boxes") inability to explain their decisions. This interpretability is vital for making maintenance decisions and validating the model's reliability, leading to improved trust and acceptance of AI-driven predictive maintenance strategies. Explainable AI is the solution because it provides human-understandable insights into how the AI model arrives at its predictions. In this regard, the paper presents Explainable AI-based predictive maintenance of Industrial rotating machines. The proposed approach unfolds in four comprehensive stages: 1) Multi-sensor based multi-fault (5 different fault classes) data acquisition; 2) frequency-domain statistical feature extraction; and c) comparison of results for multiple AI algorithms, and d) XAI integration using "Local Interpretable Model Agnostic Explanation (LIME)", "SHapley Additive exPlanation (SHAP)", "Partial Dependence Plot (PDP)" and "Individual Conditional Expectation (ICE)" to interpret the results.