Improved Imprecise Dirichlet Model–Improved Transitional Markov Chain Monte Carlo for Power System Reliability Assessment

Component outage records in power systems are often limited, which makes it difficult to represent failure probabilities with deterministic point estimates. To address this issue, this paper proposes a reliability assessment framework that combines an improved Imprecise Dirichlet Model (IDM) with improved Transitional Markov Chain Monte Carlo (iTMCMC). The improved IDM introduces a sample-size-dependent hyperparameter to construct adaptive outage-probability intervals for different equipment categories. These interval probabilities are then propagated through iTMCMC to obtain interval-valued system reliability indices. In the sampling process, a reliability-oriented likelihood function is used to guide system-state exploration, and self-normalized weights are applied to maintain estimator consistency. A case study is conducted on a standard IEEE reliability test system. The results show that the improved IDM can provide adaptive component outage-probability intervals, while iTMCMC achieves more stable LOLP and EENS estimates than MC and MCMC. The interval propagation results further demonstrate that the proposed framework can transfer component-level probability uncertainty into system-level reliability-index intervals. The proposed method provides a practical tool for reliability assessment when component failure records are incomplete or insufficient.