LAUSR

Abstract Rolling element bearings are commonly used in rotatory machines and their operation conditions need to be monitored to prevent catastrophic failures. This research aims at developing an effective online monitoring method to analyse wear severity and wear mechanisms of bearings. For this purpose, morphological features of wear debris from bearings, including quantity, colour, size and shape, are extracted from videos of moving particles carried in lubrication oil. Moving particles are detected and tracked based on the Gaussian background mixture model and the blob detection algorithm, and the target particles are separated from the image background by background subtraction. Then particle shape and dimensional features both in two- and three-dimensions are extracted from multiple images captured from different views. At the same time, the numbers of moving particles are determined based on the tracking results. The developed techniques for online particle feature extraction are applied to a rolling element bearing test rig. The particle quantity is used to indicate the trend of wear state evolution, while the colour, size and shape features of wear debris are used to classify particles into different types for wear mechanism analysis. The relationship between wear mechanisms and the trend of wear states in the time domain is established. Based on the gathered information and wear analyses, a comprehensive characterisation is obtained for online monitoring the wear state evolution as well as evaluating the wear severity and wear modes. The proposed method is able to provide wear mechanism information for root cause analysis and thus enhance the capabilities of existing online wear debris monitoring techniques.