考虑个体差异的车轮二元相关退化可靠性分析

Reliability analysis of a wheel incorporating bivariate correlation degradation considering individual disparities

由于动车组车轮受加工装配及工况载荷等不确定因素的影响,车轮退化过程复杂,存在显著的个体差异及多个性能退化特征量,且多个退化特征量间相互耦合,仅考虑单个性能特征量难以全面反映其退化过程。因此,以轮缘和轮径作为车轮退化特征量,通过随机化尺度参数表征车轮个体差异,采用Gamma过程和Copula函数建立二元相关退化模型,并根据赤池信息准则(Akaike Information Criterion, AIC)筛选Copula函数,得到可靠度解析式。基于某型动车组车轮实测磨耗数据,对车轮进行可靠性分析,同时通过车辆动力学模型进行仿真验证。结果显示:考虑二元相关时的可靠性结果比仅考虑一元退化更贴近车辆实际运行情况;根据动力学仿真结果体现了考虑个体差异的必要性,表明所提模型能够更准确地表征车轮退化过程,可为可靠性分析以及维修决策优化提供理论支撑。

Due to the influence of various uncertain factors such as processing,assembly,and working load of the Electric Multiple Unit(EMU)wheels,the degradation process of the wheels is complex.It exhibits significant individual differences and manifests multiple performance degradation features that are interrelated and coupled with each other.Therefore,considering only a single performance feature may not fully capture the intricacies of the degradation process.Hence,the characteristic parameters chosen for analysis were the wheel flange and the wheel diameter.Individual differences among wheels were represented by randomizing scale parameters.A binary correlation degradation model was then established by integrating the Gamma degradation process with copula functions.To estimate the unknown parameters,the Markov Chain Monte Carlo(MCMC)method was employed.After evaluating various copula functions,the optimal one was chosen using the Akaike Information Criterion(AIC).Subsequently,the corresponding reliability analytical formula was derived.By utilizing measured wheel wear data from a specific type of EMU,the reliability of the wheel was analyzed.Additionally,a vehicle dynamics model was constructed to validate the simulation results.The findings indicate that when reliability reaches 0.9,the estimated operational mileage under binary correlation consideration is approximately 143300 km,accounting for individual differences.Conversely,without considering individual differences,the projected operational mileage is approximately 284700 km.Notably,reliability outcomes are more conservative and closely align with actual vehicle operations when accounting for the binary correlation of the wheels,as opposed to solely considering one-dimensional degradation.The necessity of accounting for individual differences was validated through dynamic simulation results.It was demonstrated that the proposed model,which incorporates individual differences,more accurately characterizes the wheel degradation process.This provides essential theoretical support for reliability analysis,residual life estimation,and maintenance decision optimization.Such findings hold significant theoretical value and research importance in advancing the understanding and management of wheel degradation in practical applications.