Category: CBM/PHM
-
CWRU Bearing Dataset: End-to-End PHM Portfolio Project
Build a complete PHM pipeline with CWRU bearing data: fault detection, feature engineering, and RUL prediction from raw vibration signals.
-
Gearbox Fault Classification: Why Voting Beat Stacking
Soft voting beat stacking by 0.7% on real gearbox fault data. Why simpler ensembles outperform complex ones in multi-fault PHM systems.
-
SimCLR for Bearing Fault Classification with Few Labels
SimCLR pretraining hits 89.3% accuracy with only 5 labeled bearing fault samples. Step-by-step contrastive learning on CWRU dataset.
-
Self-Supervised Contrastive Learning for Fault Diagnosis
SimCLR contrastive pretraining boosted fault diagnosis from 73% to 89% using 8,000 unlabeled vibration signals. Here's the exact PHM workflow.
-
SCADA Data Integration for Multi-Asset PHM Pipelines
MQTT vs direct writes: we tested both on 47 SCADA assets. The hybrid approach cut sync lag 73% and survived 8 months in production.
-
Spectral Kurtosis for Bearing Faults: Outperforming FFT
Spectral kurtosis caught bearing failure 3 weeks before FFT showed symptoms. When it works, when it fails, and what runs in production.
-
Kalman vs Particle Filter for Bearing State Estimation
Kalman vs particle filter for bearing state estimation: real latency numbers on edge hardware, tuning tricks, and when each one breaks down.
-
Physics-Informed Neural Nets for Bearing RUL Prediction
Physics-Informed NNs cut bearing RUL error 30% vs pure deep learning. How Paris' law constraints work when you only have 20 training samples.