Abstract: The lubrication and wear failure of the sliding bearing pair in aviation fuel gear pumps is the main factor leading to gear pump failures. Traditional sliding bearing life prediction methods rely on a large amount of experimental data and have limitations such as high cost and harsh testing conditions. This article proposes a full load failure life assessment method that integrates lubrication and wear mechanism models with active learning. It comprehensively considers the effects of temperature, elastic deformation, and rough surfaces, and combines uncertainty in production and service to construct a sliding bearing lubrication and wear simulation model. The lubrication characteristics and dynamic wear behavior are characterized, and the learning and prediction process of the cumulative probability function is optimized using active learning methods, significantly reducing sample requirements and achieving efficient evaluation of failure probability and life distribution under full load conditions. In addition, the impact of various uncertainty factors on lifespan was analyzed through moment independent sensitivity analysis. Research has shown that with the increase of wear, various performance indicators of sliding bearings have deteriorated, proving the rationality of using wear as a failure criterion. On this basis, the bearing life distribution obtained through active learning methods shows that the failure of sliding bearings mainly occurs at approximately 687.88 and 859.60 hours of operation, and their total life will not exceed 900 hours. In addition, sensitivity analysis revealed that the sensitivity of the random tolerance of the bearing clearance was 0.9982, which is greater than the random pressure pulsation at the pump outlet. The service life of the sliding bearing is mainly affected by the random tolerance of the bearing clearance. The method proposed in this article for predicting and evaluating the life of sliding bearings in aviation fuel gear pumps based on lubrication and wear mechanism models will provide structural optimization efficiency for friction pairs, providing theoretical support and engineering guidance for the long-term and high reliability design of aviation fuel gear pumps.

Key words: Aviation fuel gear pump, Wear mechanism, Active learning, Life prediction, Moment independent sensitivity