Development of a Low-Cost, Micro-Electromechanical Systems-Based Sensor-Integrated Bearing for Early Fault Detection

Detecting bearing faults in their early stages can help prevent major breakdowns, reduce downtime in rotating machines. However, the conventional sensors for condition monitoring are expensive and cannot be placed near the fault due to their larger size, resulting in a longer signal transfer path that attenuates the signal and may lead to information loss. Hence, to address these challenges, the present work proposes a novel approach to enhance incipient fault detection ability by integrating a small form factor, low-cost MEMS (Micro-Electro-Mechanical Systems) accelerometer into the bearing. Experiments are conducted on an in-house test rig employing a variety of fault scenarios and operating conditions to demonstrate the effectiveness of the proposed sensor-integrated bearings. The results obtained by the MEMS accelerometer are compared with the conventional piezoelectric sensor results to assess the incipient fault detection ability of the proposed approach. Data-driven machine-learning techniques are employed to accurately classify fault signatures acquired by the MEMS and piezoelectric sensors. The results show that the MEMS sensor integrated into the bearing achieves approximately 100% fault classification accuracy. In contrast, piezoelectric and bearing housing MEMS sensors achieve 96.77% and 95.70%, respectively. The MEMS sensor integrated into the bearing successfully detects and diagnoses faults at an early stage, with approximately 70 times less cost than the conventional piezoelectric counterpart.