A numerical approach for predicting the remaining useful life of an aviation hydraulic pump based on monitoring abrasive debris generation. (February 2020)

• Record Type: Journal Article
• Authors: Li, Tongyang; Wang, Shaoping; Zio, Enrico; Shi, Jian; Ma, Zhonghai
• Is Part Of: Mechanical systems and signal processing. Volume 136(2020)
• Publication Date: 2020-02
• Subjects: Remaining useful life prediction -- Oil debris detection -- Axial piston pump; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2019.106519
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: Abrasive debris generation model with rough sliding under mixed lubrication is proposed. A partition-integration RUL prediction framework is proposed. The small sample problem of aviation piston pump is taken into consideration. Monte Carlo sampling is employed to address uncertainty. Abstract: The effective operation of aviation hydraulic pumps is directly related to the safety of passengers. Accurate remaining useful life (RUL) prediction will help establish maintenance strategy, which will reduce the risks of passengers. Currently, the degradation processes developing in the aviation hydraulic pump cannot be accurately modeled, because of the complexity of the experiment and lack of data. To address the problem, we resort to a numerical approach in this paper, taking uncertainty into consideration. Wear debris features are simulated by Monte Carlo sampling from an abrasive debris generation model. The model describes the relationship between rough surfaces and abrasive debris at the microscale level. A partition-integration RUL prediction framework is proposed to apply the microscale model to the macroscale process of pump degradation. The proposed method is validated by experimental data. The results show that the model is able to cover extreme dry contact conditions and gives good results with small amounts of data.