The performance and importance analysis of power systems based on bayesian networks

Because the power systems are becoming more gigantic, it is important for the power corporations to monitor the performance of power systems and determine which object needs maintenance most in the operation. With the advantages of describing uncertain variables and conditional independence relationships, we introduce the Bayesian network (BN) to build the performance and importance analysis model of power systems in this paper. The standard multilayer BN (MLBN) unit is put forward at first to represent different kinds of inner or outer factors in the power system. Then, the special meanings of nodes and edges in the equipment layer, station layer and network layer of MLBN are discussed in detail. Third, the integration method of MLBN in these three layers is also described to facilitate the modeling and inference process. Based on the built MLBN model of power system, the system performance and importance analysis approaches are demonstrated with corresponding posterior probability distributions. At last, the case study based on the Yunnan electric power corporation (China) is implemented. The practical transformer model shows that the proposed MLBN method can describe the inner & outer factors and relationships well to provide useful performance and importance analysis helps.