Vehicle Handling Inverse Dynamics Modeling And Simulation While Encountering An Emergency Collision Avoidance

A brief review of domestic and foreign vehicle handling inverse dynamics research is introducedin this paper. Using RBF neural networks and optimal control method, the simulation research onvehicle handling inverse dynamics in high-speed emergency avoidance with three degree-of-freedomvehicle model is analyzed. Research shows that solving the “inverse problem” not only be able toidentify the steering angel input, but also can compare the maneuverability of different vehicles. Themain purposes of this paper are:(1)A three-degree-freedom closed-loop system for steering angel input vehicle model isestablished, arrange the real vehicle serpentine test in low speed to verify the closed-loop system’svalidity. The results show that MATLAB simulation of the vehicle’s yaw rate, lateral acceleration andbody roll angel have good consistency with the actually experiment, lay the foundation of obtainingthe exact neural networks training sample.(2)Using the uniform design to arrange the closed-loop system the double lane and serpentinetest; through combining three different parameters in driver’s model to simulate different driversdriving the same vehicle to obtain the neural network’s training data, establish the neural networkmodel between vehicle’s yaw rate, lateral acceleration, body roll angel and steering angel input,angular velocity; Based on the established neural network, analyze the error between the network’sidentification and simulation results, and in the input samples to join evenly distributed random noise,further analyze the network’s accuracy and robustness.(3)Based on optimal control method, using for high-speed emergency avoidance and repeatedlyavoidance condition for vehicle handling inverse dynamics research. The steering angel input of3-DOF vehicle mode is the control variable, accurately tracking the expected path is the control object,the optimal control problem can be converted into a nonlinear programming problem while using thestate variables conversion, which can be solved by the sequential quadratic programming(SQP)algorithm. The results show that vehicle can well track the expected path in high speed, be able toidentity the steering angel input during the whole movement process, be able to compare themaneuverability of different vehicles tracking the same path in high speed, and have good consistencywith the ADAMS/CAR simulation results.In conclusion, this paper puts forward GRNN neural networks and optimal control method forhigh-speed emergency avoidance vehicle handling inverse dynamics. GRNN neural networks can identify the steering angel and angular velocity with ideal precision, high accuracy and theanti-interference ability during the entire movement process; and the optimal method not only canidentify the steering angel input, but also be able to compare the maneuverability of different vehicles,and the movement trend of simulation results is similar to that of ADAMS/CAR simulation results.The methods in this paper for solving high-speed emergency avoidance vehicle handling inversedynamics provide a new way, and for the optimization design the certain reference and theoreticalbasis.