Automated bearing fault diagnostics with cost-effective vibration sensor

Abstract

A non-continuous condition monitoring approach through periodic vibration measurement has become a common practice in the industry. However, this approach can lead to serious misinterpretation, where rapidly growing faults, that might occur in rolling element bearings, could be missed. In contrast, a continuous condition monitoring approach offers a more optimal solution in which the bearing condition is continuously tracked. This way, catastrophic failures can be anticipated in advance thus allowing an optimal maintenance action. Despite its advantages, a continuous condition monitoring is not well adopted by the industry because of the high investment cost, where the sensor cost is a major factor. To remedy this gap, cost-effective vibration sensors are therefore needed. Yet, as shown in this paper, cost-effective vibration sensors inherently comprise some technical limitation, e.g. high background noise. These shortcoming needs to be solved before these sensors can be used in an industrial setting. This paper first discusses the selection of a cost-effective vibration sensor from the market and the sensor deployment for condition monitoring purposes. Afterwards, a novel diagnostics method that can deal with the high background noise of the sensor is presented. To demonstrate the feasibility of our approach, vibration signals acquired with a high-end accelerometer and the selected cost-effective accelerometer on an industrially representative gearbox were analysed. The results show that the diagnostics performance of the cost-effective accelerometer is comparable with the one of the high-end accelerometer.