Abstract
Accurate reliability assessment of pipeline units remains challenging due to complex degradation processes and multiple sources of uncertainty. To address this issue, this paper proposes a dynamic reliability assessment model that integrates historical failure data with mechanistic principles. To overcome the computational bottleneck of conventional reliability models in handling high-dimensional complex integrals, Bayesian theory is adopted. The prior distribution, derived from expert knowledge or general empirical laws, is rigorously coupled with the likelihood function of historical failure data to construct a hierarchical parameter inference framework. For accurate parameter estimation, Markov Chain Monte Carlo (MCMC) simulation is carried out within the OpenBUGS environment, and the Brooks–Gelman–Rubin (BGR) diagnostic criterion is employed to ensure convergence and statistical validity of the computed results. The case study indicates that the integration of historical failure data effectively adjusts the expected failure time (μ) and the degree of dispersion (σ), substantially narrows the 95% confidence intervals, and reduces prediction uncertainty. The updated failure probability curve shows a clear leftward shift relative to the theoretical prediction. Based on these findings, it is recommended that the key maintenance phase be advanced from the 15th year to the 12th year, thereby achieving an optimal balance between operational safety and cost-effectiveness.