Cascaded Estimation Of Key Information For Vehicle System Based On Modular Structure

Accurate key information of vehicle system,such as vehicle motion states and tireroad friction coefficient(TRFC),is essential for safe driving.When the uncertainties appear due to coarse modelling or external interference,vehicle’s characteristics cannot be described well by the existing vehicle models.Meanwhile,variable system structure parameters and complex road situation often make it harder to get key information of vehicle system during driving process.Moreover,it is not easy to obtain tire forces accurately and robustly only through tire models.Adressing these issues,this thesis tries to improve the estimaitons of the vehicle motion states and TRFC by exploring the observation of tire forces and the estimation strategy construction of key information.The main content of this thesis is listes as follows.Firstly,the theories involving observer and Kalman filter(KF)is introduced to promote the observation performance of tire forces.In this part,an adaptive sliding mode observer is designed,and the KF is utilized to process the chattering in the observation of longitudinal tire force.In addition,an unknown input observer is employed to observe lateral tire force,and the KF is added to suppress the chattering of the observation curve.The results show that the proposed observation strategies for tire forces have high accuracy and strong robustness to variable road situation.Secondly,a cascaded strategy for estimating the vehicle motion states is constructed based on a smoothing variable structure filter(SVSF).This strategy mainly consists of the realization and optimization of multiple modules,such as the dual-track vehicle dynamics model with three-degree-of-freedom(3-DOF),the observation of longitudinal tire force,and the acquisition of lateral tire force.The SVSF controls a time-varying boundary layer anchored on the sliding mode control theory,and has strong robustness to some application conditions,such as incorrect model parameters,mismatched initial state vector,and mutated measurement.Therefore,the proposed estimation strategy has better performance.Finally,the cascaded estimation of the TRFC is realized under the framework of cubature Kalman filter(CKF)by amending the Brush tire model.Under different road conditions,the proposed estimation strategy can not perform high-precision vehicle sideslip angle,but also realize the high-performance acquisition of the TRFC.