Publication: Estudio y análisis de filtros aplicados a señales vibratorias de ejes ferroviarios para la detección de fallos
Loading...
Identifiers
Publication date
2012-09
Defense date
2012-09-24
Authors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
En el presente Proyecto de Fin de Carrera se han aplicado distintos filtros digitales a las señales de vibración de un eje de ferrocarril a escala, con el fin de determinar cuál nos lleva a un diagnóstico más adecuado. Para ello, se ha adquirido un conjunto de señales de un banco de ensayos de ejes, midiendo las vibraciones de un eje sano y un eje con defecto. Las señales han sido procesadas, por separado, mediante el filtro Smooth, la Transformada de Hilbert y la Transformada Wavelet db6, que son los tipos de filtros a analizar. Posteriormente, las señales han sido clasificadas por medio de una red neuronal artificial de reconocimiento de patrones, la cual es capaz de realizar un diagnóstico automático de la condición del sistema. A partir de la efectividad de esta clasificación se ha determinado la eficacia de cada filtrado. El estudio realizado ha determinado que el filtrado elimina una gran parte del ruido aleatorio, mejorando mucho el resultado con respecto al obtenido a partir de la señal original. Aún así, las tasas de acierto siguen siendo insuficientes para la implementación al mantenimiento de ferrocarriles. _______________________________________________________________________________________________________________
In the present Final Degree Project there have been applied different digital filters to vibration signals of a scaled railway axle in order to determine which one leads us to the most adequate diagnose. To this end, a set of signals have been acquired from an axle test bench, measuring the vibrations of a healthy axle and a defected axle. The signals have been processed, separately, with Smooth filter, Hilbert Transform and db6 Wavelet Transform, which are the filters that are meant to be analyzed. Subsequently, the signals have been classified by means of a pattern recognition artificial neural network, which is capable of performing an automatic diagnosis of the condition of the system. The efficiency of each filter has been determined from the effectiveness of the classification. The study has determined that the filtering removes much of the random noise, greatly improving the result compared to the one obtained from the original signal. Anyway, success rates remain insufficient for the implementation to railway maintenance.
In the present Final Degree Project there have been applied different digital filters to vibration signals of a scaled railway axle in order to determine which one leads us to the most adequate diagnose. To this end, a set of signals have been acquired from an axle test bench, measuring the vibrations of a healthy axle and a defected axle. The signals have been processed, separately, with Smooth filter, Hilbert Transform and db6 Wavelet Transform, which are the filters that are meant to be analyzed. Subsequently, the signals have been classified by means of a pattern recognition artificial neural network, which is capable of performing an automatic diagnosis of the condition of the system. The efficiency of each filter has been determined from the effectiveness of the classification. The study has determined that the filtering removes much of the random noise, greatly improving the result compared to the one obtained from the original signal. Anyway, success rates remain insufficient for the implementation to railway maintenance.
Description
Keywords
Tecnología ferroviaria, Ejes ferroviarios, Vibraciones, Detección de fallos