Smart sensor for electrical machine monitoring through statistical analysis

These days monitoring systems for electrical machines require high-performance instrumentation, low-cost primary sensors, and characteristics of online signal processing. Smart sensors are an approach to increase performance and overcome costs in monitoring systems. This work presents preliminary results on the implementation of a smart sensor for electrical machine monitoring integrated as a low-cost SoC. The proposed smart sensor is based on a reconfigurable hardware architecture implemented into a field programmable gate array. It extracts statistics like mean, variance, information entropy, and the distribution histogram from the acquired signal. It can work as a standalone system, showing obtained results on a VGA screen or LCD, or streaming out obtained results for further processing. Two study cases demonstrate its effectiveness during electrical machine monitoring: Detection of time-dependent harmonic effects on the voltage signal provided by a VSD. Identification of broken bars and unbalance in induction motors.