Diagnosis and Prognosis of Chassis Systems in Autonomous Driving Conditions

Expanding various future mobilities such as purpose built vehicle (PBV), urban air mobility (UAM), and robo-taxi, the application of autonomous driving system (ADS) technology is also spreading. The main point of ADS is to ensure safety by monitoring vehicle anomalies to prevent functional failure or accident. In this study, a model-based diagnosis and prognosis process was established using degradation data generated during autonomous driving simulation. A vehicle model was designed using Modelica/Dymola, and autonomous driving simulation was performed by integrating the lane keeping assistant (LKA) system with the vehicle model using Matlab/Simulink. Degradation data for the 3 components (a shock absorber damper, a suspension bush, and a tire) of the chassis system were input into the integrated simulation model. The degradation behavior was monitored with K-nearest neighbor (K-NN) and Gaussian mixture model (GMM). The remaining useful life (RUL) for each component was estimated using a Gaussian process. As a result, a normal/abnormal data classifier was designed to diagnose the autonomous vehicle simulation model, and the RUL was estimated within the 95% prediction interval.