Oxidation wear monitoring based on the color extraction of on-line wear debris

Oxidation associated wear usually involves high temperature and often accelerates lubrication degradation and failure processes. The color of oxide wear debris highly corresponds with the severities of oxidation wear. Therefore, on-line detection of oxide wear debris has the advantage of revealing the wear condition in a timely manner. This paper presents a color extraction method of wear debris for on-line oxidation monitoring. Images of moving wear particles in lubricant were captured via an on-line imaging system. Image preprocessing methods were adopted to separate wear particles from the background and to improve the image quality through a motion-blurred restoration process before the colors of the wear debris were extracted. By doing this, two typical types of oxide wear debris, red Fe2O3 and black Fe3O4, were identified. Furthermore, a statistical clustering model was established for automatic determination of the two typical types of oxide wear particles. Finally, the effectiveness of the proposed method was verified by performing real-time oxidation wear monitoring of experimental data. The proposed method provides a feasible approach to detect early oxidation wear and monitor its progress in a running machine.