Investigation on Wear Diagnosis of Aero-Engine Mechanical System Based on Lubricant Wear Particle Analysis

To improve the accuracy of mechanical system wear diagnosis in aero-engines, this paper extracts characteristic parameters such as size, color, and texture from wear particle images, constructs a correlation between wear particle characteristic parameters and wear types, and forms a typical wear particle database. Based on neural networks, an intelligent identification method for wear particle types is established, and the accuracy of wear particle identification is discussed. The results show that the identification accuracy of normal wear particles, spherical wear particles, and cutting wear particles can exceed 85%. After improvement through hierarchical, parameter addition, and multiple method fusion, the identification accuracy of fatigue wear particles and sliding wear particles has been significantly improved, and the identification accuracy can exceed 80%.