Please wait ...
0% Complete
Home
/
15th International Conference on Computer and Knowledge Engineering
TriFuse-PdM: High-Fidelity Machine Failure Prediction Using Hybrid Resampling and Model Calibration
Authors :
Saghar Shafaati
1
Javad Mohammadzadeh
2
1- Department of Computer Engineering, South Tehran Branch,Islamic Azad University, Tehran, Iran
2- Department of Computer Engineering, Ka.C., Islamic Azad University, Karaj, Iran
Keywords :
Predictive Maintenance (PdM)،Class Imbalance،Ensemble Learning،SHAP Explainability،Hybrid Resampling (SMOTE-Tomek)
Abstract :
Abstract— Predictive maintenance (PdM) is vital for modern industrial systems, enabling timely failure prediction through data-driven methods. Yet, class imbalance in failure datasets significantly hinders the accurate detection of rare but critical events. This paper presents TriFuse-PdM, a unified and interpretable PdM framework that integrates SMOTE-Tomek resampling, cost-sensitive learning, ensemble models (Random Forest, XGBoost), and a deep Multi-Layer Perceptron. SHAP is used for interpretability, and performance is assessed via ROC-AUC, PR-AUC, MCC, and calibration analysis on the AI4I 2020 dataset. TriFuse-PdM achieves high predictive accuracy (ROC-AUC ≈ 0.995, PR-AUC > 0.99) while ensuring reliable uncertainty estimation and actionable insights, offering a scalable solution for Industry 4.0 maintenance challenges. Specifically, the framework tackles class imbalance through SMOTE-Tomek resampling and cost-sensitive training, enhances interpretability via SHAP analysis, and improves prediction reliability through probability calibration.
Papers List
List of archived papers
Reliability Evaluation of 4:2 Compressors Based on Hammock Networks
Farshad Safaei - Mohammad mahdi Emadi Kouchak - Sara Talebpour
A New Hypercube Variant: Pruned Shuffle Connected Cube
Reza Latifi - Mahmoud Naghibzadeh
Analysis of Insect-plant Interactions Affected by Mining operations, A Graph Mining Approach
Mohammad Heydari - Ali Bayat - Amir Albadvi
Information Theoretic Learning-based Deep Embedded Clustering (ITL-DEC)
Hoda Shad - Mona Zamiri - Tahereh Bahreini - Reza Monsefi - Ghoshe Abed Hodtani
Realism in Action: Anomaly-Aware Diagnosis of Brain Tumors from Medical Images Using YOLOv8 and DeiT
Seyed Mohammad Hossein Hashemi - Leila Safari - Mohsen Hooshmand - Amirhossein Dadashzadeh Taromi
A Synergistic Hybrid Architecture with Residual Attention and Mixture-of-Experts for Robust Hour-Ahead Forex Forecasting
Alireza Abbaszadeh - Seyyed Abed Hosseini - Mohammad Reza Akbarzadeh Totonchi
Segmentation of Coronary Artery Stenosis in X-ray Angiography using Mamba Models
Fatemeh Fouladi - Ali Rostami - Hedieh Sajedi
PowerLinear Activation Functions with application to the first layer of CNNs
Kamyar Nasiri - Kamaledin Ghiasi-Shirazi
Improvement of CluStream Algorithm Using Sliding Window for the Clustering of Data Streams
Sahar Ahsani - Morteza Yousef Sanati - Muharram Mansoorizadeh
Distilled BERT Model In Natural Language Processing
Yazdan Zandiye Vakili - Avisa Fallah - Hedieh Sajedi
more
Samin Hamayesh - Version 46.0.2