ABSTRACT The post‐earthquake recovery of substation systems is a complex, multidimensional, and highly uncertain dynamic stochastic process. To quantify its uncertainties and enhance the flexibility and efficiency of seismic resilience assessment for substations, this study proposes a novel resilience framework, which integrates the Dynamic Bayesian Network (DBN)‐based stochastic recovery model with the Bayesian Network (BN)‐based system analytical model. Firstly, a probabilistic recovery model of the component is developed. Training data are generated according to equipment seismic fragility, expert knowledge and correlations among uncertainties, and then parameter learning is conducted to accurately evaluate the time‐varying probability distribution of component functionality. Secondly, the BN analysis model is established using a typical 220/110/35 kV substation as a case study. The functional probabilities of components are updated sequentially at each time step based on the system recovery path, producing the functional recovery curves and the resilience metrics. Finally, the effects of seismic intensities, recovery strategies, and uncertainty factors on the resilience assessment of the system are investigated. This method quantifies the uncertainties of the recovery process and evaluates the system functionality from a probabilistic perspective. Therefore, it avoids the extensive Monte–Carlo simulations required in the traditional functionality recovery assessments, enhancing the flexibility and efficiency of system resilience evaluation.
Guo et al. (Thu,) studied this question.