Research And Application Of Key Technology Of Maintenance System For Urban Rail Vehicles

Scientific maintenance is the foundation of secure and reliable operation for urban rail vehicles. With the explosive development of urban rail transportation, the demand for the maintenance is growing, and in-depth research of maintenance, based on a full knowledge of it, is needed to improve the quality of maintenance in many directions. So the frame of the maintenance management of urban rail vehicle is systematically researched in this paper, and the critical theories and methods of maintenance are enriched and applied.Firstly, the basic framework of the maintenance research for urban rail vehicle is established based on the research of equipment maintenance theories and the actual investigation from metro company. The basic contents and common means of each research direction are also summarized and analyzed.Secondly, the critical systems of urban rail vehicle are determined by using fuzzy synthetic evaluation method. Based on the failure mode analysis of each specific system, critical degree is calculated to determine the critical systems. Then the failure distribution and related reliability indicators of critical systems are determined by reliability data analysis, and the existing maintenance procedures of urban rail vehicle were comparatively analyzed to lay the root for the later study.Thirdly, the fuzzy multi-object optimization method is applied to optimize the maintenance cycle of urban rail vehicle. Taking availability, reliability and maintenance cost as the relevant targets, the single-objective optimization cycles are determined. Further, fuzzy analytical hierarchy process (FAHP) is applied to improve fuzzy multi-objective optimization model, with which a multi-objective optimization model of maintenance cycle is established and the maintenance cycle is determined.Finally, Bayesian networks are utilized to provide support for fault diagnosis of maintenance. The Bayesian network is built by the conversion of Fault tree, and the concept of diagnosis coefficient, based on diagnosis probability, time, difficulty and impact, is proposed as the basis for the optimal operation of fault diagnosis. With Bayesian network and diagnosis coefficient, the failure of door system is analyzed and the operational recommendations are proposed to make the fault diagnosis more scientific and reasonable.