Research On Vehicle Dynamic Control Based On Prediction Of The Driver Behavior Model

In recent years, Car accidents decreased year by year, but the traffic accidentcasualty rate was constantly growing. This shows that vehicle safety in China is low,and road traffic safety situation is still very bad. According to the accident statisticsreport, the vast majority of accidents are caused by the driver because of wrongoperation. Hence, the research on active safety based on the characteristics of driverbehavior is a research trend.In this paper, partly sponsored by National Science Foundation of China forYoung Scholars (51105169;51205156,51475206), Jilin Provincial Projects ofScience and Technology Development (20140204010GX), and Changchun SpecialProjects of Significant Science Research (14KG094). I put different proficiencydrivers in the driver-vehicle-road closed-loop model as the research object, andanalyzed the characteristic of driver behavior, then finished the research on vehicledynamics control of MPC based on the driver behavior.Firstly, according to the information of simulated testing based on thedriver-vehicle-road closed-loop system, I established the parameter identificationalgorithm for driver model. Secondly, on this basis, I designed a vehicle dynamicscontrol model, which use MPC to integrated control for the active front wheelsteering and directly yawing moment control, based on the characteristic of driverbehavior. The MPC aided vehicle to follow the ideal side-slip angle and idealyawing angular velocity. Finally, I set up an offline simulation platform, andsimulation test have been carried out to verify the developed algorithm.In this paper, the main content includes:1. The offline simulation test in double shift line and serpentine line.In this paper, firstly, it need to analyze the characteristic of driver behavior, so Ido experiment to get the data that was used to identify the characteristic parametersof driver behavior. Considering the workspace of controller included workingcondition of extreme danger, and ensured the accuracy of the identificationalgorithm. Choose the green hand, generally, three different types of skilled driver asthe experimental subject and double line, serpentine test conditions, meanwhile sethigh speed offline test.2. The parameter identification for ideal driver behavior model.In this paper, I choose the optimal prediction at the curvature model as the idealdriver behavior model. The parameter identification for ideal driver behavior modelincludes parameter identification for vehicle model and driver model. Firstly, I usethe least squares method to identify the second order transfer function of vehicle model. Discrete continuous transfer function and use the least squares method toidentify the parameter of discrete transfer function, and then transformed the discretetransfer function into continuous transfer function. Secondly, the parametersidentification of driver model identification based on improved genetic algorithm.3. The parameter identification for driver behavior based on neural networkIn this paper, the parameters identification of driver model identification basedon improved genetic algorithm is effective, but identification process elapsed time,and real-time performance is poor.In this paper, using neural network to connected the driver of the initial datainformation for model parameter identification with the parameter identificationresults of driver model. This algorithm can greatly improve the efficiency of thedriver model parameter identification, and ensure the effectiveness of the controller.4. The chassis dynamics control of MPC based on the driver behavior.In this paper, I used the driver behavior characteristic while driving into thevehicle stability controller. Firstly, I use the driver’s steering behavior characteristicsto predict the next moment the steering wheel angle. Secondly, I use the forecast ofthe steering wheel angle to estimate characterization of ideal vehicle steady statevariables, included the center of mass of side-slip angle and yawing angular velocity.Finally,I designed a vehicle dynamics control model, which use MPC to integratedcontrol for the active front wheel steering and directly yawing moment control,based on the steady-state deviation of side slip angle and yaw velocity.5. The offline simulation testI establish the simulation test platform included Matlab/Simulink and CarSim.Through offline simulation platform, and under test conditions to validate and debugeach algorithm, finally analyze the offline simulation results.