Abstract: In order to accurately evaluate the reliability of the main electrical connection for the sending-end converter station for Kunliulong hybrid DC project of China, the equivalent two-state models of static components, dynamic components and auxiliary components in the main electrical connection of Kunbei converter station are built. In order to reduce the computational complexity,the sending-end converter station is partitioned according to the fault consequence and location of the components. Then, the GO method and fuzzy dynamic Bayesian network are combined to built the reliability model of the system. The proposed model can directly reflect the time-sequential change process of the system reliability, and can use the posterior probability to analyze the vulnerability of the main electrical connection. The feasibility and effectiveness of the proposed model and method are proven through the analysis of simulation results.