Comparison of piezoelectric and MEMS based vibration measurement system for motors condition monitoring

Vibration analysis is a common method used in condition monitoring for rotating machinery. The purpose is to make early predictions of system failure. The analysis commonly involves a conventional piezoelectric vibration sensor. The use of these accelerometers requires high costs for vibration measurement at various points. On the other hand, there are a lot of accelerometers that potentially work well for the same purpose. Studies have found that MEMS sensors have consistent signal characteristics to piezoelectric vibration sensors. Along with the rapid development of AI, vibration analysis that used to be carried out manually by an expert has a lot of potential to be analyzed through machine learning. It opens a lot of potential to utilize sensors with limited resources. Because of that, this study aims to use MEMS accelerometers as vibration sensors. The research begins by measuring the vibration of the DC motor system using two different sensors. They are IEPE-based and MEMS-based accelerometers. Finally, Three DC motor system conditions (normal, degraded and abnormal) were measured. The data show that the MEMS accelerometer is able to differentiate the three system conditions. Besides its potential, the MEMS-based accelerometers still can be improved by further processing before it can be used as an input feature of the machine-learning model.