Sistema de diagnóstico de fallas Profibus-DP de la máquina MP5 de grupo familia mediante redes neuronales artificiales

Abstract

ABSTRACT In this research, a PROFIBUS-DP fault diagnosis system of the Family Group MP5 machine is developed using Artificial Neural Networks, which uses a model trained with the keras package, based on measurements of common faults in the physical layer that presents a PROFIBUS-DP network. The system issues a diagnosis of a probable failure based on measurements made with an oscilloscope within a PROFIBUS-DP network. The research project describes the design and implementation of a system that allows diagnosing faults in the physical layer of a PROFIBUS-DP network, with measurements made with an oscilloscope on the transmission channels of said network. The waveform indicates if there is excess cable, a short circuit or if a terminating resistor is missing in the network, the data obtained from the waveform is saved in a file generated by an application that works together with an oscilloscope, finally the file is exported to another interface that contains the previously trained classification model. The program gives a percentage of what type of failure may be present. The probability of certainty is divided into six case studies: five common failures and the correct network operation scenario.