Data Challenge
Factory Hack Korea 2024
(Top Award)
Task: CNC milling machine tool RUL prediction & optimal operating condition
Method: 1) Adaptive labeling 2) Transformer 3) Physics-based curve fitting 4) Kernel density estimation based ensemble method
POSTECH Open Innovation Big Data Challenge Competition (2023)
(Encouragement Award)
Task: Solar power generation ensemble prediction
Method: 1) Transformer 2) Kernel density estimation based ensemble method
PHM Asia Pacific Conference 2023 Data Challenge Competition
(Rank: 2nd place)
Task: Anomaly detection & Fault diagnosis for spacecraft
Method: 1) XGBoost 2) Rule-based model
2023 IEEE ICPHM Data Challenge Competition
(Rank: 1st place)
Task: Fault classification of a gear dataset
Method: 1) Inter-class maximum mean discrepancy (Metric learning) 2) Depth-wise 1-d convolutional neural network 3) Ensemble method
Presentation: 2023 / 6 / 6 (Montreal, Québec, Canada)
POSTECH Open Innovation Big Data Challenge Competition (2022)
(Excellence Prize)
Task: Solar power generation prediction
Method: 1) Weighted Forcast Observatory 2) CNN-LSTM model 3) Ensemble-based Dynamic confidence interval
PHM Society Data Challenge 2022
(Rank: 2nd place)
Task: Fault classification of a rock drill dataset
Method: 1) Physics-based cropping technique 2) Domain adaptation 3) Metric learning 4) Ensemble method
Presentation: 2022 / 11 / 08 (Nashville, Tennessee, United States of America)
Hyundai Heavy Industries Group The 2nd Shipbuilding/Marine Big Data/AI Contest for Digital Innovation (2022)
(Encouragement Award)
Task 1: Image classification
Task 2: OCR Algorithm
Method: 1-1) Imbalanced weighting 1-2) Ensemble training 1-3)Bayseign Optimization
2) Transformation / Feature extraction / Sequence (Bidirectional LSTM) / Prediction
PHM Society Data Challenge 2021
(Rank: 4th place)
Task: Predict the remaining useful lifetime (RUL) for aircraft engines under conditions of high variability
Method: 1) augmentation through sliding window method 2) 1D CNN architecture 3) ensemble method