Linear and Quadratic Time–Frequency Analysis of Vibration for Fault Detection and Identification of NFR Trains
Jyoti Kumar Barman, Durlav Hazarika
Abstract
The work reported in this article aims at developing a simple sensor-based system that detects and identifies faults of a train. Four different trains are considered for analysis as suggested by the Northeast Frontier Railway (NFR) authorities. For this purpose, an embedded system containing an ADXL335 sensor is used to capture the vibration of a railway track during the movement of a train over the track. The embedded system transfers the captured signals from ADXL335 to a laptop. These signals are processed using linear time-frequency transform (wavelet transform) in conjunction with quadratic time-frequency transform (Wigner-Ville transform) to find out whether there are any faults and thereby quantify the quality of the moving trains. Wheel-flat fault is detected for one train with the wheel position and the bogie number. This is a low-cost technique as the method involves only one ADXL335 sensor and an Arduino development board, and the software used for the analysis is python that is an open-source and platform-independent software.