Research On Maintenance Strategy Of Freight Train Wheel Based On Partially Observable Markov Decision Process

The rapid expansion of railway freight transport capacity and the transformation of freight train maintenance strategy from Planning Maintenance to Condition-Based maintenance have brought great challenges to the maintenance management of freight trains.As a key component of freight trains,whether wheels run safely and reliably will directly affect the operation quality and transportation efficiency of freight trains.At present,there is a problem in the detection process of railway freight train wheels that the quality of the detection data is not high enough to accurately reflect the true state of the wheels.Therefore,it is very important to formulate a maintenance strategy that conforms to the current status of wheel detection.During the operation of the wheels,both the tread and diameter will wear and degrade,which will threaten the safety and stability of railway freight trains.In order to restore the normal tread pattern of the wheels and ensure the safe operation of the freight trains,the worn wheels must be reprofiled or replaced.However,repairs will result in a reduction in wheel diameter and an increase in maintenance costs.With more recovery of the freight train wheels,the wheels will be scrapped prematurely.With less recovery,the risk of tread wear will increase.Therefore,"when to repair and how to repair" has become a key issue to be solved urgently in the formulation of railway freight train wheel maintenance strategy.In view of the partially observable characteristics of the real wheel status and the time continuous characteristics of the wheel monitoring data,a dynamic maintenance strategy based on the Partially Observable Markov Decision Process(POMDP)model is proposed in order to reduce wheel maintenance Cost and extend the service life of wheels.Firstly,on the basis of analyzing the characteristics of wheel tread degradation data,a mixed effect model of tread wear degradation is constructed,and the law of diameter wear is determined by the correlation between the diameter wear and tread wear.Secondly,by collecting the data of the diameter repair amount and the value of current income diameter,the relationship between the two is fitted to establish the degradation rule of the wheel diameter during the reprofiling process.Then,using the conditional degradation amount derived from the degradation law of tread and diameter as an important input,the degradation process of wheel tread and diameter is dispersed into a finite degradation state,and a two-dimensional state space is established.With the object of minimizing long-run expected maintenance cost per unit time,the maintenancestrategy model based on POMDP is established and a policy-iteration algorithm is used to determine the optimal maintenance strategy.Finally,based on the construction of the above degradation model and maintenance strategy model,according to the measured data of freight train wheels,the maintenance strategy proposed in this thesis is compared with the strategy of age model and periodic inspection model to verify the superiority of this maintenance strategy.