Tire Force Estimation Using Model-Based and Machine Learning Approaches for Distributed Drive Electric Vehicle

Distributed drive electric vehicles are one of the key focuses in the contemporary automotive industry, providing greater flexibility for active safety control. The realization of vehicle control relies on accurate vehicle driving state information. Precise estimation of tire forces is essential, since tire forces directly influence vehicle dynamics. This paper focuses on distributed drive electric vehicles and proposes two estimation methods: model-based method and machine learning method. The model-based estimation method employs PID observers and Unscented Kalman Filters to estimate longitudinal and lateral tire forces respectively. The machine learning estimation method utilizes Long Short-Term Memory networks for tire forces estimation. The effectiveness and performance of two methods are verified by MATLAB/CarSim. The results demonstrate that the machine learning method provides more accurate tire force estimation and exhibits robustness against measurement noise.