基于模糊BN和改进证据理论的车辆故障定位方法

Vehicle Fault Location Method Based on Fuzzy BN and Improved Evidence Theory

为解决车辆售后维修过程中由于车辆部件的关联故障而产生大量杂乱故障码,导致通过分析故障码进行源头故障部件定位困难的问题,本文提出一种基于模糊BN和改进证据理论的车辆故障定位方法。首先,根据历史数据及专家经验构建模糊BN模型并得到其后验概率。其次,将后验概率作为改进证据理论的基本概率赋值输入,提出融合邓熵和Pignistic概率距离的修正系数对证据修正,解决证据不确定性及证据间冲突问题。然后,采用基于矩阵分析的证据合成规则,避免大量证据合成失败情况的同时减少计算量,得到故障定位结果。最后,以ABS系统为例,验证该方法的可行性。本文所提出的方法可为维修人员提供指导进行快速定位故障。

This paper proposes a vehicle fault location method based on fuzzy BN and improved evidence theory,in order to solve the problem of difficulty in locating the source fault parts by analyzing the fault codes due to generation of a large number of chaotic diagnosis trouble codes by the associated faults of vehicle parts in the pro‐cess of vehicle after-sales maintenance.Firstly,the fuzzy BN model is constructed according to historical data and expert experience and the posterior probability is obtained.Secondly,the posterior probability is input as the basic probability assignment of the improved evidence theory,and the correction coefficient combining Deng entropy and Pignistic probability distance is proposed to correct the evidence,so as to solve the problems of the uncertainty of the evidence itself and the conflict between evidences.Then,the evidence synthesis rule based on matrix analysis is adopted to avoid a large number of evidence synthesis failures and reduce the amount of computation,and the fault location results are obtained.Finally,the ABS system is taken as an example to verify the feasibility of this method,which can provide guidance for maintenance personnel to locate faults quickly.