Study On Modular Vehicle State Estimation

With the rise of the number of automobiles,lead to frequent occurrence of traffic accidents,people put forward higher requirements on automotive active safety performance.Therefore,a variety of vehicle dynamics stability electronic control system which can effectively improve the vehicle's active safety performance have been developed.However,in order to make these electronic control system to play a role,the vehicle state information should be obtained in real time.But due to the constraints of the cost of production and measurement error,some of the vehicle states information can not be measured directly through the vehicle sensors,which limits the improvement of the vehicle's active safety performance technology.With the continuous development of the estimation theory,researchers are more inclined to use the sensor measurement information of mass-produced vehicles,through the estimated method to obtain the vehicle state information which can not be measured.For most of the advanced automotive electronic control system,the required vehicle state information is numerous.Among them,accurate known vehicle tire force and vehicle speed information is the key of vehicle control,which also makes the problem of vehicle tire forces and vehicle velocities estimation have important theoretical significance and engineering application value.In view of the problem of vehicle tire forces and the vehicle velocities can't be measured by sensors,combined with the R&D requirement of the National Natural Science Foundation "Research on nonlinear integration estimation method of vehicle states and road parameters" which researched by our laboratory,this paper mainly studied the estimation problem of coupled vehicle states information,including estimation of vehicle longitudinal,lateral tire forces,vehicle longitudinal,lateral velocities,yaw rate and sideslip angle.Aiming at the problem of complex structure of vehicle system and strong coupling between the driving state,considering that the estimation of a certain state does not meet the requirements of the actual vehicle system,this paper propose an modular vehicle states estimation scheme.In view of the coupled vehicle states information including,including estimation of vehicle longitudinal,lateral tire forces,vehicle longitudinal,lateral velocities,yaw rate and sideslip angle,the longitudinal and lateral tire force observer modules based on the sliding mode observer theory are designed,the longitudinal velocity,lateral velocity and yaw rate observer modules based on the nonlinear full-dimensional observer theory are designed,sideslip angle observer module based on the geometrical relationship between the sideslip angle and longitudinal,lateral velocity are designed.Furthermore,according to the coupling relationship between the states,the observer modules are integrated into an modular estimation scheme.Considering that adjusting the observer gain manually involves a huge workload,and it is difficult to adjust the optimal parameters,this paper propose an optimization method of observer parameters based on genetic algorithm.The parameters of longitudinal and lateral tire force observer and yaw rate observer are optimized by standard genetic algorithm.The multi-objective genetic algorithm is used to optimize the parameters of longitudinal and lateral velocity observer whose estimation results influence each other.In order to verify the estimation effect of the modular vehicle state observer designed in this paper,the vehicle simulation experiment is needed.Considering the safety and economic factors,this paper uses the high fidelity dynamic simulation software ve DYNA to carry out the simulation experiment verification.Firstly,the four-wheel drive vehicle model in ve DYNA is used as the simulation vehicle,and the estimation effect of the tire force sliding mode observer is verified,and the Extended Kalman Filtering(EKF)estimation method is compared and analyzed.Then two typical conditions are selected,and the results of the modular vehicle state observer are analyzed and verified.Finally,the HQ430 vehicle model built in ve DYNA is used as the simulation vehicle,three kinds of real vehicle driving conditions are selected,with HQ430 real vehicle measurement informations as inputs of observer,the estimations of vehicle state from observer are compared and analyzed with the measured value.The simulation results show that the modular vehicle state observer designed in this paper,has estimation accuracy in the case of off-line simulation and real vehicle data verification.