Physics-informed failure mechanism for hydraulic pumps: state-of-the-art review, challenges and trends

Prognostic and health management (PHM) of hydraulic pump systems plays a vital role in reducing component failure, and ensuring the security, reliability as well as the practicality of various industrial applications. Traditional methods rely on physics-based models or data-driven learning for PHM, however, the accuracy and availability of these approaches are always limited by intrinsic challenges such as the physical interpretability of diagnostic results and root causes of failure induction in terms of complex hydraulic systems, which has received little attention. This work attempts to summarize and review recent research with a comprehensive overview of physics-informed failure mechanism (PIFM) for PHM analysis of hydraulic pumps in terms of fault evolution process and failure physical laws. Specifically, we start with a brief structural explanation of several commonly used hydraulic pumps and their working principles, and then the typical failures such as mechanical- and hydraulic failures of pumps, are classified systematically. Furtherly, the pump failure mechanism caused by materials, micro-evolution of the friction interface and macroscopic friction under internal excitation are reviewed. Meanwhile, failure mechanisms affected by external single excitation, external multiple excitation and internal-external hybrid excitation are also summarized and discussed. Eventually, some potential challenges and strategies related to fault evolution, degradation mechanism and physics-informed machine learning (PIML) are presented. This comprehensive survey is expected to pay more attention to enhancing intelligent maintenance strategies, operational reliability and decision-making of hydraulic pumps.