Reliability Analysis Of Power Supply System Of Metros Signal

Signal equipment plays a very important role in urban rail transit system.Different types of computers,switches and other high-quality precision electronic equipment need stable power supply.Once the signal power supply system has problems,it may lead to train outage,or even endanger the safety of running.At present,the main reliability analysis methods applied in the field of urban rail transit signal power supply system are Fault Mode and Effect Analysis(FMEA),Fault Tree Analysis(FTA)and Monte Carlo method.Because of the large number of internal units and the complex operation state of the signal power supply system,when applying the above-mentioned methods,the FMEA method is subjectively affected and may neglect the role of low-risk factors;the FTA method is too workload and error-prone to evaluate the reliability of complex systems;the Monte Carlo method can not get a high-precision solution,and there are random errors.None of the above methods can be used to analyze the dynamic model of complex systems,the correlation among internal units and the polymorphism among units.At the same time,due to the inherent shortcomings of the analytical method,when more components are considered,the amount of calculation is very large.Therefore,a few units in the signal power supply system are not considered,and the reliability model is relatively simple.The key components such as buses,terminals and switches are not taken into account.Combining the advantages of GO method and Bayesian method,GO-Bayesian method is proposed.This method has the GO method's characteristic of convenient to model and powerful reasoning ability of Bayesian network.It also omits the processing link of common signals in GO method,simplifies calculation,and considers various states of the system In this thesis,GO-Bayesian method is used to analyze the availability curve and steady-state availability indexs of signal power supply system in urban rail transit with different UPS redundancy conditions.The posterior probability of each unit in the system with each redundancy mode is analyzed,and the weak links in the system are analyzed so as to facilitate the key checking and maintenance in daily maintenance.The innovations of this thesis are as follows:(1)The transformation algorithm from GO graph to dynamic Bayesian network is proposed,the structure and CPTs of nodes of operators in GO Method in dynamic Bayesian network is studied.(2)Ignoring the correlation among the units will inevitably lead to inaccurate model.Considering the correlation among the components in the series structure of urban rail transit signal power supply system,the structure related to shutdown is equivalent to a Bayesian network node,and the calculation formula of the conditional probability table of the node is obtained by studying its state transition diagram.(3)A new GO Method operator,type 18 operator,is added,and the algorithm of transforming the operator to dynamic Bayesian network is studied.(4)Common cause failure is a very important cause of system failure.Based on dynamic Bayesian network,dies the modeling method of series-parallel structure with common cause failure is studied.The analysis shows that the method can be used to model the system intuitively and conveniently,and has advantages in calculation speed and accuracy.The calculation shows that the result of the optimized model is in line with the actual working conditions.There are 60 figures,35 tables and 52 references in this thesis.