Explainable Predictive Maintenance Using Random Forest and SHAP for Industrial Machine Failure Prediction

This work addresses a solution for Explainable Predictive Maintenance (XPdM) combining machine learning algorithms and explainable AI methods for industry. The Random Forest has an effective prediction accuracy, and SHAP gives us an explicit and human-understandable explanation. The developed dashboard with Streamlit shows these explanations to the user in real time.   The XPdM solution presented here is capable of overcoming three major problems: imbalance classes, interpretability, and on-line application. The key factors identified are tool wear, temperature, and rotational speed, providing better decisions to improve maintenance efficiency. In essence, the XPdM system closes the gap between complex ML models and industrial use cases.    

  [1] “AI4I 2020 Predictive Maintenance Dataset,” UCI Machine Learning Repository, 2020. [2] L. Breiman, “Random forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001. [3] S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” in Proc. Advances in Neural Information Processing Systems (NeurIPS), 2017. [4] McKinsey & Company, “The future of predictive maintenance,” 2018. [5] World Economic Forum, “Industry 4.0: The fourth industrial revolution,” 2019. [6] “Scikit-learn: Machine learning in Python – Documentation,” 2023. [7] “SHAP (SHapley Additive exPlanations) – Documentation,” 2023. [8] “Streamlit: The fastest way to build data apps – Documentation,” 2023. [9] A. K. S. Jardine, D. Lin, and D. Banjevic, “A review on machinery diagnostics and prognostics implementing condition-based maintenance,” Mechanical Systems and Signal Processing, 2006. [10] T. Chen and C. Guestrin, “XGBoost: A scalable tree boosting system,” in Proc. ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining (KDD), 2016.