Recognition of defects in railroad wheels from patterns in acoustic signatures
Freight car wheels are subjected to wear leading to mechanical defects which can subsequently result in derailments. The purpose of railroading is the safe, rapid and economical movement of material. The most frequently occurring defects are cracks on the tread, flange, rim and plate, flat spots, etc. It is believed that these various defects can be detected using acoustic signature inspection. Various algorithms have been developed which detect some of the above mentioned defects but so far there is no algorithm developed which can detect all the defects. A data base of acoustic signatures was established from a set of wheels whose nature was known. Some of the previously developed algorithms were tested on this set of acoustic signatures. A new algorithm was developed using the time dependence nature of the acoustic signature. This algorithm had a better detection rate than any of the other algorithms developed for the detection of defective wheels. Also, the feasibility of time domain analysis of the acoustic signature using Hilbert transformation was examined, for the detection of defective railroad wheels.