Vehicle State Estimation Based on Multidimensional Information Fusion

Abstract

As the core input parameters of various control systems, the real-time and accurate acquisition of reference speed, mass and road slope is the key factor to improve the performance of intelligent vehicle dynamics control. Therefore, the parameter estimation method based on multi-dimensional information fusion is proposed in this paper. A comprehensive evaluation of wheel dynamics state is realized by information fusion, which is quantified in terms of wheel speed credibility. Then the calculated dynamic speed and auxiliary speed are weighted coupled to achieve accurate estimation of reference speed, which avoids the influence of unstable wheels. Similarly, the method to calculate the confidence factor of mass estimation is established in order to screen the vehicle state suitable for estimation. And the online estimation of mass is realized based on recursive least square method. Meanwhile, the road slope estimation algorithm based on interactive multiple model has been designed, which achieves the weighted fusion of Kalman filter observer based on kinematics and unscented Kalman filter observer based on dynamics. Finally, the road tests were carried out on representative working conditions. The maximum error between the actual speed and the reference speed does not exceed 0.68m/s. The relative error of mass estimation is not more than 1.95%, and the absolute error of slope estimation is less than 1.84%, which proves that the proposed estimation algorithm has high comprehensive performance. More importantly, it is not limited to specific working conditions, which means a great significance for the development of intelligent vehicles.