Modelamiento basado en datos para la detección de fallos en una bomba hidráulica centrífuga multietapa vertical a partir de señales multimodales en condición normal

Abstract

This work details the development process of a new methodology for the creation of fault detectors based on the use of generative models obtained from multimodal signals of a machine operating in normal state. Using for this purpose a GAN, and a one-class classifier (OC-SVM).