Wear-based time-variant reliability analysis on variable geometry turbine stator

Variable geometry turbine (VGT) stators are critical for enhancing operational flexibility and efficiency of aerospace engines, yet traditional time-invariant reliability analysis ignores the wear-induced performance degradation of adjusting mechanism. This study establishes an integrated wear-based time-variant reliability analysis framework for VGT stators, which innovatively reconstructs the physical coupling of Archard theory and multi-body dynamics (MBD) for multi-kinematic pairs, and customarily integrates the LHS-BP neural network with time-variant limit-state functions and MTBF conversion. First, multi-source uncertainties (load, dimension, material) are quantified with statistical and interval methods. Second, an MBD-embedded Archard wear model is built to quantify the 1000-cycle nonlinear wear evolution of six key kinematic pairs. Third, time-dependent limit-state functions are developed for three failure modes: excessive adjusting error, insufficient driving force, and unbearable applied stress. Finally, the LHS-BP framework is adopted for reliability evaluation and MTBF conversion, realizing dynamic reliability tracking with operating cycles. This work fills the research gap in VGT stator wear-induced time-variant reliability analysis, clarifies its performance degradation law, and provides a scientific basis for reliability design and life prediction.