Research On Fault Diagnosis Of 750kV Substation Based On Bayesian Network

750 kV substation system is the kernel of 750 kV northwest power grid,its reliable operation has important influence on stability of power grid and power supply quality. With the scale expansion of northwest power grid, number increasment of substation, complicated circuit structure, requiring increasment of the protection configuration and the complexity of network structure, 750 k V substation fault diagnosis requirements is further improved.When the 750 kV substation failure occurs, operators need to obtain effective and reliable information from a lot of gathering information in a short period of time, and make accurate decisions by processing them, locate fault components, isolate fault zone quickly, and then try to analyze the cause of the problem in time to realise the grid quick recovery. The intellectualized development of substation makes the study of intelligent substation fault diagnosis technology is particularly important.Based on the lines complexity and redundancy configuration characteristics of 750 kV substation system, by analysing the uncertainty and diversity characteristics of fault information and consider the omission or error and other cases during the processing of information, this thesis put forward the 750 kV substation fault diagnosis based on bayesian network method. This method made full use of the alarm information, protection and circuit breaker status information, element action sequence information and fault wave record information collected by supervisory control and data acquisition system, established the main and redancy network model of suspicious fault components based on bayesian network, descriped the causal logic between the associated node, the initial probability of the bayesian network diagnosis model is assigned using information-entropy theory, the temporal characteristics of sequence of event information is used to correct diagnosis model, and extracted fault features of fault wave record information using wavelet analysis, evaluated the motion behavior of the element to determine the fault components, to establish fusion diagnosis model, the advantages of fault wave record information diagnostic methods and sequential fault diagnosis method are fused at decisional level of Dempster-Shafer(DS) Cevidential theory, and then according to the synthetic rules to judge the fault. The simulation results show that this method have higher precision and better effectiveness.