Friction characteristic prediction assisted by intelligent wear state identification of valve plate in aerospace highly compact servo motor pumps

The valve plate pair is a critical friction pair in aerospace servo motor-pumps, whose wear failure under extreme high-pressure and high-speed conditions severely restricts equipment reliability and service life. Conventional research methods depend heavily on extensive life tests and user feedback, suffering from long cycles, high costs, and inability to quantitatively characterize multi-mode wear or observe microscopic wear evolution in real time. This study establishes a 1:1 scale dedicated friction and wear test rig replicating actual service conditions, systematically investigating the tribological properties of typical cylinder block/valve plate material combinations. A friction characteristic prediction model coupling hydrodynamic lubrication and multi-body dynamics is developed, addressing the limitation of traditional pure hydrodynamic models in low-speed boundary lubrication regions, achieving a maximum prediction error of 12.2% under stable rotational speed. Furthermore, a YOLO-based instance segmentation method is proposed to automatically identify and quantify abrasive scratches and adhesive wear pits, overcoming the subjectivity and inefficiency of manual observation. This method provides finer-grained validation data for Archard wear models by distinguishing independent contributions of different wear modes. The results show that DLC (diamond-like carbon) coatings effectively isolate direct metal contact and significantly inhibit adhesive wear. This work provides theoretical and experimental support for material optimization and life prediction of servo motor-pump valve plate pairs.