Reliability Assessmnt Of The Tration Substation Based On Bayesian Network

With the rapid development of China’s railway mileage, the continuous improvement of the share of electrified railway,the continuous increase of railway transport capacity and the continuous improvement of the runnning speed, traction power supply system, as a core part of the electrified railway, which carries out to provide power for the electric locomotive, the reliability of which is directly related to the security and the stability of railway transportation;at the same time,tration substation is the bridge between catenary and external power system, whose reliability is directly related to the level of the reliability of the entire electric railway. Thus, the evaluation of the reliability of railway substation is a very important theoretical and practical significance to assure a safe, stable and reliable operation of railway substation.Bayesian network (BN) method is used to analyze and evaluate the reliability on traction substation. Combined with the structural features of traction substation and the basic principle of Bayesian network, first of all, according to the difference of reliability block diagram on the analysis of the target, different types of Bayesian networks are built, such as model which is based on fault tree, minimal path and minimal cut,at the same time, which is analyzed by combining with the concrete instance. Taking a typical traction substation for example, according to the four definitions under fault conditions that are the single feeder faults, singer-arm failure, up(or down) failure and the entire traction substation. Four Bayesian network models of main electrical connection of traction substation are established under the fours definitions, the models are diagnosed, their reliability are anaylzed and evaluated, the weak links which influence system reliability are finds out and their reliability indexes are calculated. The quantitative and qualitative assessment of reliability is realized.Secondly, the whole reliability of the computer protection device and traction substation integrated automation system are evaluated, combined with the hierarchy of the computer protection device and traction substation integrated automation system, the Bayesian network models of them are established and calculated, the main effect factors are gotten,their weak links are found out. Bayesian network model not only can make the forward and backward cause and effect inference but also can make qualitative and quantitative analysis. Thus, it is a practical way to traction substation reliability research.Finally, Fuzzy Analytic Hierarchy Process (FAHP) is used in the reliability allocation about traction substation main electrical connection,and also Bayesian network is used in the reliability allocation about the spacer layer unit of the traction substation integrated automation system. FAHP not only take into account a variety of factors that affect the reliability allocation, such as complexity, important dgree, technical level, repair factors, working environment, cost and expense, and so on, but also take into account the uncertain factors of the proportion when the experts compare the two elements in the judgment matrix, the consistency checking of the Analytic Hierarchy Process is eliminated. The reliability allocation results are basically identical with engineering Practice, which shows that the method is more effective and practical.Making full use of BN’s advantage that backward reasoning can be caused by a system failure, the reliability indexes about the spacer layer unit of the traction substation integrated automation system are assigned by calculating the conditional probability of failure of the components when the system fails.