Model-based vehicle dynamics control system and states estimation for 4WD Formula SAE electric vehicle.

Yaw Control is a common topic on high performance race cars vehicle dynamics. Compared to Yaw Control for passenger cars, the application to high-performance vehicles is mainly focused on improving drivability, reactivity and performance through the partial or total achievement of a desired cornering behaviour, obtained through vehicle states tracking. The subject of this analysis is the development of a Yaw Control for the Formula Student (FS) 4-Wheel-Drive (4WD) electric race car of Squadra Corse PoliTo. The prototype is equipped with four electric motors, independently controlled, driving one wheel each, guaranteeing high flexibility in torque control. Thus, in this application, Yaw Control is obtained exploiting Torque Vectoring (TV) strategy, so obtaining a torque unbalance between left and right wheels during cornering, to help the vehicle behave in the desired way. The desired cornering vehicle behaviour is decided generating reference signals for Yaw Rate and Side Slip Angle, two fundamental states for lateral vehicle dynamics. Yaw Rate reference is compared to measured Yaw Rate, while Side Slip Angle reference is compared to an estimated Side Slip Angle, since measuring that state is not suitable for the application, apart from using very expensive sensors. For this reason, an EKF-based combined with a kinematic-based Side Slip Angle estimator has been decided to be implemented, to have a reliable state estimation. This work aims to propose, adapt, and implement a solution to improve vehicle performance during transient and steady state manoeuvres, through the adoption of a combination of a LQR and model-based Feed Forward TV Yaw Control strategy. This solution has been decided to be used, considering actual state of the art Yaw Control strategies, considering ease of implementation, accuracy and computational effort to run in an ECU. Both the state estimator and the control logic are implemented and tested in a Matlab/Simulink and Vi-CarRealTime co-simulation environment. The goodness of the results are showed analysing cost function trends, typical performance-related vehicle signals and lap-time on Formula Student Skidpad Event. Comparative analysis are carried out between controlled and uncontrolled vehicles, both on standard automotive manoeuvres like constant radius cornering, ramp steer and sine sweep steer, and on real FS event. Moreover, the controller, together with the state estimator, has been deployed on the Squadra Corse PoliTo prototype’s ECU (dSpace MicroAutobox II) for track testing and validation. Results show a reduction of Skidpad lap-time of 5%.