Reliability Modeling and Algorithm Implementation of Ship Dynamic Spare Systems Based on Dynamic Bayesian Networks

Objectives To address the low efficiency and poor accuracy in constructing Conditional Probability Tables (CPT) for the reliability modeling of shipborne dynamic spare systems, this paper proposes an automatic CPT generation method based on spare gate structures and state combination deduction. Methods A Bayesian Network (BN) is established by mapping nodes and dependencies from the fault tree. The probability distribution of nodes is automatically derived from the input states of spare gates, thereby enabling automated CPT generation. In addition, a cross-time-slice dependency mechanism is introduced to characterize the activation and repair processes of standby units, leading to the construction of a Dynamic Bayesian Network (DBN) model. Results The results of CPT automatic generation show that the proposed method requires 5.341 seconds of computation and outperforms the traditional manual assignment approach in terms of accuracy and robustness. It effectively addresses the excessive workload and high error rates in CPT construction under complex redundancy configurations. Conclusions The proposed method significantly reduces the complexity of DBN modeling, improves the precision and efficiency of reliability analysis, and provides an efficient and standardized tool for the reliability modeling and quantitative evaluation of ship spare systems, offering promising application prospects.