Intelligent Detection and Classification System for Rail Surface Defects Based on Eddy Current
Description
This work presents a system to detect squats on rails based on a device that monitors the train wheel's oscillation frequency. In it, the sensor should be installed directly on the service trains. It uses wavelet power Spectrum in Frequencies between 1060-1160 Hz and a simple threshold for Squat identification. This work is limited to squats detection for preventive maintenance.
Files
Steps to reproduce
Signal samples csv files ( 1000 points / signal) - Weld, Squat and Joint Each sign corresponds to a row of csv files, the last column refers to a category indicator number. The signals were acquired in rails and later filtered so that they were highlighted and stored according to the category, after this step the wavelet transform of each signal was performed to feed the CNN network input and elaborate the proposed classification model.