Direct Tire Slip Angle Estimation Using Intelligent Tire Equipped With PVDF Sensors

Tire slip angle is a vital parameter in vehicle dynamics control. This article proposes a tire slip angle estimation method using intelligent tire technology and machine learning algorithms. First, a finite element model (FEM) of an intelligent tire with five polyvinylidene fluoride (PVDF) piezoelectric film sensors (PVDF sensor) attached to the inner liner is built using ABAQUS software, and the validity of the FEM is verified. Then, the signal responses of sensors in different slip angle, load, tire pressure, vehicle speed, slip ratio, and tread wear under the tire rolling state are analyzed. Subsequently, promising input feature values are selected based on signal response analysis and dimensionally reduced using the principal component analysis theory. Finally, these feature values are used for training three different machine learning techniques with the purpose of slip angle online estimation. The test results show that the slip angle estimation model using Gaussian process regression has the best performance, with a tenfold cross-validation mean error of 5.47%. At the same time, the estimation model also can accurately estimate the tire slip angle in extreme maneuvers, with the mean absolute percentage error of 7.16%. In general, the proposed slip angle estimation method is feasible and has great potential to improve the vehicle motion control performance especially in extreme maneuvers.