Data-Driven Tire-Road Friction Estimation for Electric-Wheel Vehicle With Data Category Selection and Uncertainty Evaluation

Abstract

The tire-road friction coefficient (TRFC) is a key parameter for precise motion control of electric-wheel vehicle. In the article, a data-driven method is proposed, focusing on multidomain feature fusion to achieve TRFC estimation under longitudinal maneuvering signal excitation. The longitudinal maneuvers encompass both the stationary maneuvers (e.g., longitudinal acceleration is around 0) and nonstationary maneuvers (e.g., all maneuvers except stationary ones). First, a scheme is introduced for analyzing vehicle states and parameters related to TRFC, which can provide the data category selection for the data-driven method. Moreover, a stochastic variational deep kernel learning framework is devised to efficiently map spatial–temporal features and assess estimation uncertainty. Subsequently, the TRFC estimation method is validated through simulations on various road surfaces. Results demonstrate the superiority of the proposed method over classical techniques, including a data-driven method and a model-driven method. Furthermore, experimental validation confirms the effectiveness of the proposed method.