High Speed Railway Traction Power Supply Equipment Maintenance Strategy Considering State Evolution Process

The traction power supply equipment is very important for high-speed railways.With the application of Prognostics and Health Management(PHM)theory and Condition-based Maintenance(CBM)theory,maintenance manage of traction power supply equipment is in urgent need of upgrading.In view of the current requirements in the traction power supply system,in this paper the equipment’s health status deterioration process model was established based on Gauss-Poisson process,and the maintenance decision-making framework was set up based on Markov Decision Process(MDP).The partially observable character in equipment’s health status information and the time continuous character in deterioration process were considered.Then the maintenance decision-making models based on Partially Observable Markov Decision Processes(POMDP)and Partially Observable Semi-Markov Decision Processes(POSMDP)were established.Further,a multi-objective optimization model was established to weigh the balance between reliability and economy,and a Multi-objective Particle Swarm Optimization(MOPSO)algorithm was adopted to solve the model.In the PHM-CBM framework of traction power supply equipment,this papers puts forward a health status composite degradation process model based on Gauss-Poisson process to reflect continuous deterioration and sudden deterioration factors at the same time.Gauss-Poisson health degeneration model is used to divide the traction power supply equipment’s health status and calculate the state transition probability matrixes.A traction power supply equipment maintenance decision model was established based on MDP using the state transition probability matrixes.This paper considers the fuzziness and uncertainty existed in the process of equipment health status evaluation.The observation matrix and the belief status space are used to describe the state observation probability space,and a POMDP model was established for traction power supply equipment maintenance plan’s formulation.TheOne-pass algorithm is applied to compute the model,and a traction power supply equipment maintenance scheme is established considering the ambiguity of equipment health status.This paper considers the time continuity of the equipment health status degradation process,and the semi-markov core is used to describe the equipment state transition probability with continuous time.This improved the performance of the discrete MDP model,and a more flexible maintenance plan decision-making model was established based on POSMDP,so more kinds of status distribution functions apply to the model.Persues algorithm is used for carrying out the annual maintenance plan.The requirement of multi-objective optimization for maintenance decision was considered,the Expected Time of Power Not Supplied(ETPNS)was used to represent the maintenance’s influence on reliability.Then a multi-objective optimization model based on equipment’s reliability and economy is proposed,and a Multi-objective Particle Swarm Optimization algorithm for this model is designed.This method achieves a better balance between reliable and economical.The main contribution of this work is the study of the realization of the entire procedures of traction power supply equipment’s condition-based maintenance policy,which involves degradation modeling,maintenance decision modeling,multi-objective optimization decision modeling and it’s solving.This work can serve as references for PHM-CBM in maintenance decision.The model can solve the problems of health state evaluation uncertainty and time continuity in condition-based maintenance,and provide reference for multi-objective optimization decision modeling.The research work in the dissertation achieves optimal annual maintenance scheduling,the results show that the model is feasible and effective,providing basis to implement scientific and reasonable condition-based maintenance schemes for the traction power supply equipment.