Embedded Bearing Fault Detection Platform Design for the Drivetrain System in the Future Industry 4.0 Era

As industry 4.0 becomes more and more prevalent, predictive maintenance (PdM) systems are gradually being valued by the industry. To reduce the risk of overall system failure which is caused by faulty components. The real-time monitoring sensors must collect the sensing data continuously, and the subsequent computing system can determine the categories of faulty components in the target machinery. The drivetrain system is critical because it drives the operation of the overall motor system in factories. Bearings are one of the important components in the drivetrain system. A healthy bearing can reduce friction and make the shaft rod in the drivetrain system work smoothly. Accelerometers are usually used to collect the vibration signals for further fault detection in machinery. However, accelerometers are expensive and consumables that they need to be replaced frequently. Moreover, using accelerometers to collect bearings vibration signals is also restricted by the operating temperature, humidity, ground loops, and the rest. This paper used a three-axis vibration sensor to detect the accelerated vibration signals of faulty bearings. Then, the Hilbert transform is employed to determine the spectral envelope of a waveform, which leverages the bearing fault detection with the random forest algorithm. By integrating the vibration sensor module and the fault detection computing module, the proposed embedded fault detection platform is proper for the goal of high-accuracy and low-cost bearing fault detection for the drivetrain system.