Bearing Health Monitoring

Health monitoring of bearings is a widely researched topic and has been attempted by analysing acoustic, thermal and vibration signatures. The methods usually require signal of a healthy bearing to be used as a baseline. This limits their use in practical scenarios. This work proposes a kurtosis based baseline free method of analysing vibration signals to identify the bearing which has generated a fault. It then reports a detailed study on empirical mode decomposition technique for extracting intrinsic mode functions and suggests a set of steps which are necessary and sufficient for the purpose of bearing health monitoring. Thereafter, it compares a few dominant frequencies with the expected ones based on known bearing dimensions. This process has been shown to be fairly accurate in identifying the location of fault in a bearing.