Fault diagnosis of a vehicle with soft computing methods
Chapter in Scopus
- Additional Document Info
- View All
The main goals of an on-board fault detection and diagnosis system in a vehicle are mainly to avoid damage to the vehicle and prevent dangerous situations for occupants. These goals can achieved by triggering correcting actions or at least by warning the driver. However, a vehicle is a complex system where monitoring and diagnosis is challenging due to inherent uncertainty caused by noisy sensors, measurements and non modeled dynamics. In this work a new approach is presented to perform online diagnosis in a vehicle, which is able to deal with measurement noise presence. Our method analyzes with fuzzy logic and neural networks the residuals obtained by the comparison between sensors measurements and the output of a mathematical model of vehicle dynamics. Experiments shows promising results when faults are induced during different vehicle maneuvers simulations. © 2008 Springer Berlin Heidelberg.