Due the high level of control and automation networks in modern industries, the sensor-based systems play a key role in the industrial scenario. Actual predictive techniques commonly use sensors to anticipate machinery malfunctions, this kind of intervention has a high operational value once it can avoid corrective maintenance stops, i. e. before the failure reaches a high level of severity and compromise the machine. In consequence of that, the development of sensors applied to non-destructive techniques (NDT) for failure monitoring in electrical machines have become a recurrent theme in recent studies. In this paper, it was employed the vibration analysis technique, which is a NDT that already proved to be efficient for detection of many structural anomalies in induction motors. Besides that, it was shown an alternative approach for the costly commercial sensors using the piezoelectric low-cost sensor, the feasible market prices can turn this NDT more likely to industrial applications. In order to verify the capability of piezoelectric sensor as a transducer for vibration analysis, its frequency response was performed using the pencil lead break (PLB) test. After this validation, the RMS value from the voltage samples obtained in the test bench was used as a signal processing method. The comparison between the results for different levels of mechanical load attached to the machine shaft indicates not only a successful performance of the low cost sensors for load estimation purposes, but also showed that oversized motors may present higher vibration levels in some components that could cause mechanical wearing.