Research on complex multi-state distribution system reliability modeling and assessment methods based on Bayesian networks and random set theory

For the difficulties of quantitative analysis in the reliability of power distribution system composed of the multi-state components and the limitations of the traditional fault-tree reliability evaluation in the distribution system,a new method of modeling and assessment of multi-state distribution system reliability based on BN and random set theory are proposed in this paper. By using the advantages of the establishment of a clear hierarchy of power distribution system model composed of the uncertain sources and figurative direct expression of Bayesian network, This model not only plays a role in distribution system reliability evaluation indicators, but also quantitatively analyses the impact of the components on the system which is on the basis of unified random set model of classical combination rules included DemPster combination rule?The model is presented in this paper provide a solid foundation for improved in distribution system. This paper uses Bayesian network to analyze a 10KV substation which contains 8 outlet, 12 circuit breaker, isolating switch 38, 106 Fuse on TAIZHOU city, Zhejiang Province. Algorithm calculate the distribution system reliability targets, but also can calculate the probability of the various components under the conditions of the probability of the system to normal or fault operational status. Improve distribution system reliability purposes can be achieved by locate to improve the reliability of certain components Therefore, the proposed algorithm can calculate the reliability index at the same time calculate the various components probability on the conditions of the system state, both draw on the overall assessment of the entire distribution system, but also play a role in diagnosis of weak links. Have been tested to verify the legitimacy and effectiveness of the algorithm, this method can be applied to a complex distribution system reliability evaluation.