In many industrial settings, a significant amount of data is generated from electro-mechanical systems and stored, without processing to gain valuable insights that could enable optimised production while bringing down maintenance cost to its barest minimum. Data mining techniques offer potential solutions to address this concern. In this paper, anomaly detection techniques using machine learning models such as K-Nearest Neighbour (KNN), Support Vector Regression (SVR) and Random Forest (RF) have been applied to vibration sensor data for early fault detection of industrial electric motors. The models relied on vibration data collected from sensors mounted on four bearings. Initial results suggest that the RF model outperformed SVR and KNN for the data set analysed, and can be a candidate data mining technology to implement for condition monitoring of electro-mechanical systems.
|Proceedings of the World Conference on Smart Trends in Systems, Security and Sustainability, WS4 2020
|2020 Fourth World Conference on Smart Trends in Systems Security and Sustainability (WorldS4) (Online)
|27/07/20 → 28/07/20
- Vibration Analysis
- condition monitoring
- anomaly detection
- Fault Detection