Predictive Maintenance in Electrical Machines: An Edge Computing Approach

Predictive maintenance is one of the premises of the Industry 4.0. A condition monitoring and fault diagnostic never seen before are possible with the new technologies embracing the digitalisation. This paper presents an edge computing approach for the predictive maintenance of electrical machines based on motor current signature analysis (MCSA), using Fast Fourier Transform (FFT) and a Support Vector Machine (SVM) classifier. A broken rotor bars and eccentricity fault detection are implemented on Arduino platform, providing a low-cost solution affordable for small and medium enterprises. The system has been tested in a lathe in one of the laboratories of the Mining and Industrial School of Almaden (UCLM) with satisfactory results.