Reliability Allocation For Rail Vehicles Based On Credibility Theory

In the field of rail vehicle maintenance, the research of reliability allocation is still in its infancy stage. Scientific and reasonable reliability allocation method is the guidance to the allocation of human and material resources in maintenance works, and the result of reliability allocation can be fed back to the manufacturer, thereby guiding the research and development of the rail vehicle products.The credibility theory is combined with traditional scoring reliability allocation method by this thesis, and the scoring allocation method based on credibility theory is proposed. The case study of reliability allocation is carried out based on the examples of 4 types of typical rail vehicles. Firstly, the credibility theory which is based on a complete axiomatic basis is introduced, overcoming the deficiency of the traditional fuzzy theory, and promising more objective results. Secondly, whole-vehicle-level and subsystem-level reliability models of the HXD3B electric locomotive, the HXN5 diesel locomotive, a type B subway train and a linear motor subway train are built. Thirdly, the scoring system is programmed based on the MATLAB GUI designing module. Fourthly, scoring standards based on triangle fuzzy numbers are designed. From the fuzzy scores given by specialists to the subsystems of the 4 types of rail vehicles mentioned above, an overall score of each subsystem is calculated using fuzzy simulation algorithm. Finally, the failure rate of the whole vehicle is allocated according to the overall score of each system, and the reliability is calculated using relative formulae. Based on the reliability allocation result, another reliability allocation is carried out among the 3 systems with the highest reliability requirement in each type of rail vehicle.Based on the comparison and the analysis of the result between the traditional scoring allocation methods and the scoring allocation method based on the credibility theory, the maximum difference is within 9%, and the average difference is around 4%. The consideration of the fuzziness has guaranteed that the result of the scoring allocation method based on credibility theory are more reliable.