Enhanced embedding surrogate modeling method of interactive components reliability evaluation for nose wheel steering mechanism

To ensure the continuous safety of the nose wheel steering mechanism (NWSM), it is essential to validate the reliability evaluation of interactive components. There are two challenges for efficient reliability evaluation: (i) the optimality of diverse limit state functions cannot be guaranteed due to the simultaneous extraction of multiple response samples; (ii) the errors of correlation measurement for interactive components are accumulated resulting from the failure hierarchy. To address these problems, this paper presents an enhanced embedding surrogate modeling method (EESM) by synergistically integrating the enhanced feedback surrogate model (EFSM) with embedding correlation (EC) theory. EFSM is applied to iteratively approximate the fitted limit state function to the real situation by quantifying and minimizing indicated errors level through sample screening mechanism. EC alleviates error accumulation in layered failure modeling by transforms the probability space of failure variables while preserving high-dimensional dependency structures. Taking the Nose Wheel Steering Mechanism (NWSM) as a case study, the analytical results of EESM by comparing with other reliability evaluation methods are presented as that (i) the best performance in modeling accuracy and efficiency is held by EESM, respectively at 0.3085, 4.9419 and 301 in MAE, R2 and training calls; (ii) EESM holds the superior simulation performance in calculational time and reliability, separately at 2.6809 s and 0.9746. The effort of this work promotes the improvement of reliability analysis for engineering systems with interactive components and multi-failure modes.