Physical‐Guided Virtual Data Generation Method and Its Application in STWT Cathode Reliability Assessment

To address the critical challenges of insufficient reliability modeling accuracy and large extrapolation bias caused by missing key stress combination data in dual‐stress accelerated degradation tests, this paper proposes a physical‐guided virtual data generation method. A nonlinear Wiener process is adopted as the degradation model, and the acceleration factor consistency principle is introduced to achieve cooperative extrapolation of drift and diffusion parameters. The Arrhenius law and the inverse power law are integrated to construct a dual‐stress acceleration relationship. A staged closed‐form estimation procedure and hierarchical Bootstrap method are employed for parameter estimation and full‐chain uncertainty propagation. A two‐level Monte Carlo algorithm generates virtual trajectories with stochastic fluctuations. The method is applied to generate virtual degradation data for the missing stress combination (1020°C, 2.5 A/cm 2 ) of space traveling wave tube cathodes. The results demonstrate that this method can generate multi‐stress reliability data with consistent physical mechanisms and credible statistical properties small sample conditions, providing a novel technical pathway for the accurate reliability assessment of high‐reliability, long‐life products.