Automotive ECUs fault diagnosis modeling based on the fault database

Abstract

The automotive ECUs is becoming more and more complicated, and so is the fault diagnosis. In order to improve the maintenance quality and efficiency, the paper proposes a fault diagnosis approach based on fault database. By making full use of data stream, we firstly extract symptom vector by processing data steam and pre-processing rules, and then we use the symptom vector to match the fault pattern in fault database, we use the unmatched vector as the test case of C4.5 decision tree algorithm to create the link rules between fault symptom and fault reason, and finally store the rules into the fault database. An example of ETCs is showed to testify the fault diagnosis method. The test result confirm the reliability and validity of this diagnosis method. © 2013 Springer-Verlag.