This study addresses the issues of accuracy and efficiency in train wheel defect detection by proposing a method based on ROCKET and LightGBM. Using wheel pressure time-series data obtained from sensors, the ROCKET algorithm is employed for time-series feature extraction. Combined with the LightGBM model and Bayesian optimization techniques, an effective defect detection method is established. Experimental results validate that this method significantly improves detection accuracy, providing robust technical support for railway transportation safety.
Enhancing Train Wheel Defect Detection with ROCKET Feature Extraction and LightGBM
2024-07-26
1175896 byte
Conference paper
Electronic Resource
English
Train positioning system based on sleeper defect feature detection
European Patent Office | 2020
|A METHOD FOR DETECTION OF WHEEL FLATTEN DEFECT ON A MOVING TRAIN
European Patent Office | 2018
|Engineering Index Backfile | 1937
Train wheel and railway turnout matched with train wheel
European Patent Office | 2020
|