Research On The Vehicle State And Road Tire Friction Coefficient Estimation Based On Dual Extended Kalman Filter

As we all know that safety, economy, and environmentalism are the three basic factors of the development of the modern vehicles. Kinds of active safety control system have made great progress, almost including each direction of the vehicle movement, in order to enhance the first factor, as well as the most important factor-safety. All these systems decide the control logic to realize the active safety based on the actual vehicle state and road friction information. So how to receive the vehicle state and the road friction coefficient actually becomes the chief problem that nowadays the vehicle safety control system has to face first. Testing the vehicle state and road friction information through the sensor has its own disadvantage that the cost is high and easily affected by environment outside. To find a actual method of receiving the vehicle state and road friction coefficient values using the ready-made sensors in vehicle, the researchers put forward various estimation method, of which have advantages and disadvantages. Based on the above, this paper applies the Extended Kalman Filter (EKF) to estimate the vehicle state and road friction at the same time, making the actual estimation come true.A single EKF estimating the vehicle state based on a 4-degree of freedom (4-DOF) vehicle model has been established primarily in the paper. The 4-DOF vehicle model describes the longitudinal, lateral, yaw, and roll four direction movement of vehicle, and the other state such as slip rate and slip angel can be calculated through these four states. Through the 80km/h double lane change test and 70km/h slalom test verifying the algorithm, the estimation values of vehicle state and simulator values from CarSim shows the great coherence, and under the situation of setting reasonable initial value, the algorithm having quick convergence. The HSRI tire model is applied to calculate the tire forces to ensure the characteristic of real time and actuality, and its estimation values have been verified by CarSim. The HSRI tire model that this paper adopt split the difference between MF tire model needing more parameters but having higher actuality and the linear tire model having simple structure but less actual. The estimation of tire forces of ESP system from BOSCH company just made use of the HSRI tire model, which means that this tire model has perfect application in reality.The vehicle response has reacted the information of tire and road, and it is a effective way to receive the road friction through vehicle state estimation values. This paper utilizes the EKF again to estimate the road friction coefficient value through the sensor test values of vehicle movement which are also parts of vehicle response based on the vehicle state estimation of last chapter. To solve the problem of Jacobian matrix calculation during the EKF design, the HSRI tire model is transformed in the structure that makes the expression show the obvious relation between tire forces and road friction coefficient. Through the steering test containing 60km/h steering test and 70km/h slalom test, and the breaking test 100km/h initial velocity breaking system, verifying the algorithm, high friction and low friction at the same time, the estimation values of road friction coefficient and road setting values in CarSim shows the great coherence. The high friction road estimation values has slower convergence than the low friction road, and the reason possibly contains in that when the vehicle at large steering or slipping such as breaking on low friction road, the algorithm is inspirited greatly, and the characteristic caters to the active safety control system acting near the limit danger situation.The vehicle state estimation design need the road friction coefficient know, and single EKF can only use the stable constant road friction values, while the road situation will change any time in real driving. So a Dual Extended Kalman Filter (DEKF) is applied to estimate and vehicle state and road friction value at the same time, which brings the changed information of road friction to the vehicle state estimation. The DEKF contains two parts of state estimator and parameters estimator, and both have two parts of prediction and correction. The two estimators change information with each other and correct the each other's prediction synchronously, to realize the actual estimation of the vehicle. The breaking test on joint road verification shows that the DEKF algorithm can actually estimate the vehicle state and road friction value on various road, but a single EKF with the situation of constant road friction value receives the estimation values discrepant widely. The convergence of road friction estimation is slower than the vehicle state estimation value, and through adjusting the measurement noise covariance matrix of the two estimator, the effect of convergence speed difference can be avoided.The comparing vehicle from CarSim is a B Class Hatchback vehicle, with no ABS. The vehicle model and tire model used in this paper both consult the parameters of the vehicle. The slip and slip angle coefficient of the HSRI tire model is obtained through s ?Fz andα?Fz relationship curve by interpolation, to ensure the precision of the tire model.