Research On Hazard Evaluation System Of Subway Running Gear Based On Reliability Analysis

Subway has many advantages and becomes an inevitable choice for large cities to solve traffic congestion problems in China.Running gear is an important part of subway,it can ensure subway running and stopping safely.If hazards which occured before faults could be grasped,effective measures could be taken to prevent hazards deterioration.The hazard evaluation system of subway running gear based on reliability analyzed on-the-spot data and grasped the regularity of hazard.The appropriate indexes were choosed according to the influence of the key parts on running gear.The hazard evaluation system can support maintenance,fault prevention and dispatching.The following researches were studied.1.Based on on-the-spot data,the hazard pattern set of subway running gear was summarized,and the key components with more hazards were extracted.In view of the existence of irreeargular truncation of on-the-spot data,this paper analyzes the hazard distribution model of the key components with more hazards in the running gear by using the survival analysis theory.The time between hazards and the occurrence rate of hazard were calculated by parameter estimation,and the regularity of hazard was grasped from on-the-spot data.2.Reliability analysis of fuzzy Bayesian networks based on fault tree.According to the hazard pattern set and the key components,the fault tree model of running gear was established.The fault tree of running gear was transformed into Bayesian network,which solves the problem that the key components of running part obey different distribution models and had more uncertain informations.Fuzzy theory was introduced into Bayesian network model,aiming at some key components which had no accurate value due to lack of on-the-spot data.The fuzzy ordered weighted averaging operator algorithm and group decision were used to aggregate the trigonometric fuzzy numbers provided by different experts.These accurate values were the prior probability of Bayesian network.The posterior probability of each node was obtained though simulation,the possibility of key parts causing hazard of running gear was ordered,and the weak links of the running part are obtained.3.According to the hazard pattern set,the importance degree of the key components and the selection principle of the hazard evaluation system of the running gear,a multi-level evaluation system of hazard of running gear was constructed.The system consisted three levels:the target layer,the criterion layer and the index layer.Taking rolling bearing as an example,the hazard indexes which based on vibration signal were extracted by MEEMD.The direct time domain parameters were extracted from the IMF component.The selection of indexes were based on the sensitivity to hazard.According to the expert suggestion and related references,the hazard grade is divided into three levels.Fuzzy ordered weighted average operator algorithm was used to calculate the threshold value given by different experts,thus the hazard interval was determined.4.The extension theory and the extension cloud model were proposed to assess hazard.The cloud model solves the shortcomings of the extension theory.The game theory was used to calculate the combination of subjective and objective weights.GAHP was proposed,which combines the advantage of GM and AHP.The extension theory and the extension cloud model integrated hazard assessment were caried out by four sets of data.The results showed that both methods could accurately assess hazard state of the running gear.The extension cloud model was more reasonable.The eigenvalues of hazard level variables could judge the evaluation level more clearly and intuitively.5.In order to evaluate the future state of hazard,a GCPSO-SVR hybrid algorithm was proposed.The basic principles and steps of SVR,PSO-SVR and GCPSO-SVR were introduced in detail.The three models were used to predect the the wheel size simulate.The results shown that the GCPSO-SVR hybrid algorithm had higher convergent and prediction effect than the other two methods,and the superiority of GCPSO-SVR hybrid algorithm was verified.According to the existing problems and the demand of the development strategy,the hazard evaluating system of running gear was constructed,which improved active prevention ability of subway.