Tire Pressure Monitoring using Weighted Horizontal Visibility Graphs

Abstract

Tire pressure monitoring systems have been proven to reduce fuel consumption and increase driver safety. Today, direct measuring systems are installed in the car, which measure the tire pressure with sensors inside the tire, or indirect systems, which detect a drop in tire pressure through the relative change in the wheel speed. This work proposes a novel way of detecting tire pressure conditions by transforming the vibration data of chassis components into a weighted horizontal visibility graph. Graph features are extracted from this representation to serve as input to an XGBoost classifier. Drives on a test track with tri-axial accelerometers on the upper control arm with low and normal tire pressure are performed to evaluate the method. The results indicate that the proposed method classifies the reduced tire pressure with high precision while also allowing changes in tire pressure to be detected quickly.