An extended Bayesian melding method for system reliability assessment considering multi-level accelerated life information

To improve the accuracy of system storage reliability assessment under small-sample conditions, this paper proposes an accelerated-test Bayesian melding method (AT-BMM) for integrating multi-level accelerated life test information. The proposed method establishes a unified probabilistic framework that jointly incorporates component prior distributions, system natural distribution, and system accelerated life test information. To address the complexity of posterior inference in AT-BMM, an importance sampling scheme with updated weights is introduced to enhance computational efficiency and estimation accuracy simultaneously. Based on simulation studies, the influence of the pooling weight parameter on posterior distribution estimation is investigated, and comparative analyses demonstrate that the proposed method achieves higher assessment accuracy than existing methods. Furthermore, the effect of uncertainty in component prior specifications is examined, showing that the proposed method improves the robustness of system-level reliability assessment. Finally, a real case study is presented to illustrate the effectiveness and practical applicability of the proposed AT-BMM.