 |
International Society of Science and Applied Technologies |
| AI-Driven Predictive Maintenance: Bayesian Reliability Assessment and XGBoost for Smart Industrial Systems | ||||
| Author |
Kuo-Yi Lin
|
|||
| Co-Author(s) |
Yu-He Hong; Meng-Hua Li; Ming-Lang Tseng
|
|||
| Abstract |
Predictive Maintenance (PdM) enhances industrial reliability by mitigating unexpected failures and optimizing resources. This study proposes a PdM framework that combines XGBoost-based failure time prediction with Bayesian reliability analysis, allowing dynamic updates of failure probabilities using real-time sensor data. The integration improves Remaining Useful Life (RUL) estimation, risk assessment, and decision support. Evaluated on the PHM08 dataset, the proposed model outperforms Random Forest and LSTM in predictive accuracy, uncertainty quantification, and interpretability. SHAP analysis reveals key factors influencing equipment degradation, increasing model transparency and stakeholder trust. Aligned with Sustainable Development Goal 9 (Industry, Innovation and Infrastructure), this approach supports smart manufacturing by enabling reliable, explainable, and adaptive maintenance strategies. It fosters resilient industrial systems through reduced downtime, better maintenance planning, and efficient decision-making. The results highlight the model’s scalability and practical value in advancing intelligent, sustainable industrial operations.
|
|||
| Keywords | Predictive Maintenance (PdM), Remaining Useful Life (RUL), XGBoost, SHAP, Bayesian Reliability, Industry 4.0, Industry, Innovation and Infrastructure | |||
| Article #: RQD2025-42 | ||||
Proceedings of 30th ISSAT International Conference on Reliability & Quality in Design |