Desenvolvimento de um identificador de falhas para motor trifásico de indução baseado em máquinas de vetores de suporte implementado em cloud

Abstract

As the Industry 4.0 advances, the induction motor sensoring becomes remote and connected to the Internet. With this, a data flow can be continuously analyzed through cloud computing in data centers. The predictive maintenance can then use this big data to increase your ability do detect faults when compared to the classical multi-signal approach. The purpose of this work is to develop a failure identifier in induction motor W22 IR3, based on current and vibration analysis. The data from these currents and vibrations are generated utilizing an embedded system for data acquisition consisting in a MEMS (Microelectromechanical Systems) controlled by a SoC (System on a Chip) and a current transformer. The data analysis will be implemented in IBM Cloud through Watson Studio and SPSS Modeler to apply an MVS statistical model that will be trained and tested using different kernel functions, using the database generated by the embedded acquisition system.