Event-driven Data Mining Techniques for Automotive Fault Diagnosis

Chaitanya Sankavaram, Anuradha Kodali, Diego Fernando Martinez Ayala, Krishna Pattipati, Satnam Singh, and Pulak Bandyopadhyay
Submission Type: 
Full Paper
phmc_10_109.pdf743.06 KBOctober 10, 2010 - 4:52pm

The increasing sophistication of electronics in vehicular systems is providing the necessary information to perform data-driven diagnostics. Specifically, the advances in automobiles enable periodic acquisition of data from telematics services and the associated dealer diagnostic data from vehicles; this requires a data-driven framework that can detect component degradations and isolate the root causes of failures. The event-driven data consists of diagnostic trouble codes (DTCs), the concomitant parameter identifiers (PIDs), customer complaints (CCs), and labor codes (LCs) associated with the repair. In this paper, we discuss a systematic data-driven diagnostic framework featuring data pre-processing, data visualization, clustering, classification, and fusion techniques and apply it to field failure datasets. The results demonstrated that the support vector machine (SVM) classifier with DTCs and customer complaints as features provide the best accuracy (74.3%) compared to any other classifier and a tree-structured classifier with SVM as the base classifier at each node achieves approximately 75.2% diagnostic accuracy. We also show that a sequential repair plan based on this classifier results in savings of about 15.7% per claim over a knowledge management approach that makes repair decisions based on observed failure frequencies only. We also illustrate the limitations of data-driven approaches, and emphasize the need for building fault models prior to vehicle release and for incrementally evolving these models with the observed cases to handle unanticipated faults in order to maximize vehicle availability.

Publication Control Number: 
Submission Keywords: 
fault diagnosis
diagnostic algorithm
Submitted by: 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed