Maintenance Policies And Robust Optimization Research On Single Component Of Urban Rail Transit Vehicle

Repair and maintenance of urban rail transit vehicles is the basis to ensureurban rail transit vehicles operating safely and quickly. Considering the detectionuncertainty and the cognitive uncertainty on the influence of urban rail transitvehicle component maintenance policy, this paper puts forward a maintenancepolicy and a robust optimization model based on a constrained Hurwicz Criterion.This helps to optimize the safety and reliability of vehicle operation, improveresource utilization efficiency, reduce the maintenance cost which has importantpractical significance and research value.Firstly, according to rail transit RAMS management specification, this paperreference the parameters of reliability, availability, maintainability, and safety,evaluate key degree of rail transit vehicle components based on RAMS, and thenclassifies rail transit vehicle components and chooses maintenance policy.Secondly, to consider the influence of uncertainty of inspection to thecomponent which suits condition based maintenance, this paper usesmeasurement-error models to get the probabilities relations between the true statesand measure states of the component, and uncertainty of inspection means themeasure error. And then calculate the distribution of the true state, using iterationway. As a result, the maintenance policy model based on LMDP is built. So, wecould get the different polices and maintenance cost under different standarddeviations of measurement by solving the model.Then, when the standard deviation of measurement is zero, the research usesrobust optimization to solve the problem of the uncertainty of cognition inparameters of Markov decision process. There are three parts of the robustoptimization model. Firstly, uncertainty level is to express the probability to get awrong transition probability matrix. Secondly, the research uses optimism level ofHurwicz criterion to decide the preference between the MAXIMIN robust versionand MAXIMAX version to get an acceptable maintenance cost. Thirdly, this paperuses a new decision variable m to make sure the transitions that are impossible in real life are never considered in robust optimization.Finally, we apply the models of choosing of maintenance policy, maintenancepolicy based on LMDP, and robust optimization to the bogie of rail transit vehicle ofa metro-transportation corporation in the city of CC. The conclusion shows thatusing LMDP and robust optimization to consider the influence of the uncertaintyand cognition will make sure that with an accepted cost we can get an optimal policyto meet the safety and reliability request of maintenance.