Evaluating An Interactive Troubleshooting System for Heavy Vehicles with Limited Historical Data

Abstract: Today’s trucks are built to be modular in order to fit the customer’s specific needs.This means that no two vehicles are  alike, and that makes it difficult to create universal models for diagnosis. When a truckor bus suffers from a breakdown, it is important that the vehicle comes back on the road as soon as possible since breakdowns cause downtime and repair costs. In a shared effort between researchers from Scania, SICS and LiU a tool forcost-efficient off-site guided troubleshooting is being developed, named DiaGuide2. This tool has an interactive interface for use both when the vehicle is in the workshop and on the road. DiaGuide2 will isolate the probable causes then recommend and plan the necessary actions, based on models that describe the vehicle, to resolve all present problems. This thesis introduces an implementation of a diagnosis tool based on a method from literature, called Fault Tree Analysis(FTA). The DiaGuide2 diagnoser module, which uses a probabilistic Bayesian Network model, will be compared against this FTA top-down deductive model. The main goal of the comparison is to find out if it is possible for the DiaGuide2 to carryout a successful diagnostic of a complex system in the vehicle, and how well it can do that when compared to the Fault Tree Analysis tool. The results of this study will helpfind out the advantages and limitations of both tools.