Intelligent Tire Slip Angle Estimation Based on Sensor Targeting Configuration

Authors:
Publisher website
Google Scholar
Abstract
Tire slip angle is a vital parameter in tire/vehicle dynamics analysis and control. This paper proposes a novel tire slip angle estimation method based on intelligent tire technology and machine learning algorithms. Firstly, the finite element modeling approach for intelligent tires with several PVDF piezoelectric film sensors (PVDF sensor) attached to the inner line is delineated, and the fidelity of this modeling approach is substantiated through actual vehicle experiments. Then, using the global sensitivity analysis method, the sensitivity of slip angle variations to the output voltage of the PVDF sensor within the potential sensor installation positions across a single tire pitch is analyzed during tire operation. On this basis, highly sensitive areas are selected as installation positions for PVDF sensors to collect richer sensor signals. Subsequently, promising input feature values are selected, and are dimensionally reduced using principal component analysis theory. Finally, these input feature values are used for training three different machine learning techniques with the purpose of tire slip angle estimation. The results reveal that the Gaussian process regression (GPR) estimation model exhibits superior overall network performance, yielding minimum and maximum mean absolute percentage errors (MAPE) of 2.01% and 5.02%, respectively, during 10-fold cross-validation, with a mean MAPE of 3.32%. Furthermore, this model demonstrates effective tire slip angle estimation even during extreme maneuvers while successfully mitigating interference arising from variations in slip ratio and tread wear.
Funding Information
  • National Natural Science Foundation of China (U20A20331, 52225212, 52172367)
  • Jiangsu Innovation Support Plan/Hong Kong, Macau and Taiwan Science and Technology Cooperation Project (BZ2022055)

This publication has 36 references indexed in Scilit: