A Study Of Vehicle Parameters Identification Based On Kalman Filter And Non-linear Observers

With the development of the traffic and the increasingly pursuit of the vehicle safety fromthe users, researchers now pay more intention to the safety problem of vehicles. TheAnti-Block System (ABS) came out firstly, afrer that, the Traction-Control System (TCS) andElectronic-Stability Program(ESP) had applied one after another. These systems are designedto control the important state variables of the running vehicles, e.g. the yaw moment and CGslip angle. They are both indirect parameters to the sensors which are integrated in thevehicles. Thus the measurements of these parameters are important. In this paper, there arethree primary parts.Firstly, a multi-body dynamics model of a civil car was built in ADAMS/View. Thevehicle front and rear suspension, power train system, vehicle body system, steering systemand other subsystems were introduced. The curves of the simulation results of the multi-bodydynamics model would be compared to the curves estimation results of the linear Kalmanfilter, to validate the precision of the filter.Secondly, considering the complexity of the observer, a2d-o-f dynamics model was builtbased on the mechanical analysis of the vehicle, which was simulated in MATLAB/Simulink.Then, a linear Kalman filter was set based on the2d-o-f model, and its parameters wasmodified.Thirdly, a non-linear double track model is built. Experimental results show that theaccuracy of the model is capable for the observer. On basis of this model, the observer basedon a linearization of the vehicle model around the currently estimated state vector is set.Lastly, based on the built multi-body dynamical model and the Kalman filter, thedifferences between the simulations and the estimations of the yaw moment and the CG slipangle are observed, as the experiment was a steer angle drive. Result turns out that theprecision of the Kalman filter is acceptable.