Thesis CONSTRUCCIÓN DE MODELOS PARA EL MONITOREO Y PRONÓSTICO DE FALLAS DE POLINES EN CORREAS TRANSPORTADORAS
Loading...
Date
2017-01
Authors
Journal Title
Journal ISSN
Volume Title
Program
DEPARTAMENTO DE INFORMÁTICA. INGENIERÍA CIVIL INFORMÁTICA
Campus
Campus San Joaquín, Santiago
Abstract
La detección automática de fallas de polines en correas transportadoras de mineral, es un problema desafiante con un alto impacto en la confiabilidad y seguridad de la operación en la gran minería a cielo abierto. Las Redes Neuronales Artificiales (ANNs) han sido utilizadas para el monitoreo de la condición
de maquinaria rotatoria durante años; en comparación, las Máquinas de Soporte
Vectorial (SVMs) son un desarrollo más reciente en el campo del monitoreo
de condición. Este trabajo examina el rendimiento de ambos clasificadores para
el monitoreo de condición de polines de correas transportadoras, usando atributos
extraídos desde señales de vibración y emisión acústica. Si bien se obtienen
buenos resultados con ambas técnicas, SVM supera a ANN en exactitud.
Automatic failure detection of conveyor rollers for mining is a challenging problem of high impact on reliability and security of mine operation. Artificial Neural Networks (ANNs) has been used for condition monitoring of rotating machinery for years; in comparison, Support Vector Machines (SVMs) are a more recent development in the condition monitoring arena. This work examines the performance of both classifiers for condition monitoring of conveyor belt rollers, using features extracted from vibration and acoustic emission signals. Although good results are obtained with both techniques, SVMoutperforms ANN accuracy.
Automatic failure detection of conveyor rollers for mining is a challenging problem of high impact on reliability and security of mine operation. Artificial Neural Networks (ANNs) has been used for condition monitoring of rotating machinery for years; in comparison, Support Vector Machines (SVMs) are a more recent development in the condition monitoring arena. This work examines the performance of both classifiers for condition monitoring of conveyor belt rollers, using features extracted from vibration and acoustic emission signals. Although good results are obtained with both techniques, SVMoutperforms ANN accuracy.
Description
Keywords
MINERÍA A TAJO ABIERTO, CORREAS TRANSPORTADORAS, MAQUINAS DE SOPORTE VECTORIAL