Research On The Coordinating.Driver Preference Cooperative Adaptive Cruise Control(CACC) Systems

As a key component of intelligent transportation system(ITS),research on Cooperative Adaptive Cruise Control(CACC)system has received the widespread attention in recent years.Personalized driver assistant system is integrated driver characteristics based on traditional assistant control system,and can further improve the utilization and acceptance of the system.However,the current research is seldom on personalized CACC system.This thesis study on the method about coordination between driver's preference and CACC's performance from the concept of human-machine coordination team.This thesis is to do research on the two key technologies:driver car-following preference online learning and coordination optimizing control is studied based on the automotive system dynamics,Kalman Filter theory and Adaptive Dynamic Programming theory.At first the personalized CACC model is established after introducing the key technology research status of CACC at home and abroad.Then the framework of the driver characteristics learning coordination control system and CACC vehicle longitudinal dynamic model are established.In view of the noise and error reduce the accuracy of evaluation algorithm,which exist in collected state data when car-follwing process,the driver preference online self-learning algorithm is established based on Kalman Filter theory.The steady following state is defined.An appropriate driver model is selected to establish a state equation which is used in Kalman Filter model.The self-learning algorithm procedure is designed.The simulation test show that the designed self-learning algorithm accurately estimate the driver's preference.In view of the contradiction between driver preference and the performance of CACC system,a blending ratio is proposed to define time headway,and a coordination control is integrated in the existing CACC dynamic model.The car-following performance indexes named security,transport efficiency,riding comfort and fuel economy are established for designing performance index function.The function is used to design the key network of Supervised Learning Adaptive Dynamic Programming model.Through optimizing the blending ratio in real time,the coordination between driver preference and the performance of CACC system will be reached.The modeling and Simulation of the controller are researched in the light of the previous design in the Matlab software environment.The effectiveness and adaptivity of the algorithm are analyzed in detail.Through the comparative simulation trial with non-coordination controller in CACC,the designed algorithm can guarantee all performance indexes into constraint range,demonstrating a good effectiveness of the algorithm.Simultaneously another simulation trial is operated that demonstrating a good adaptivity of the algorithm when driver's preference change.