| Accurate longitudinal control of the entrance and exit process of unmanned city electric buses can improve the transportation efficiency of the whole bus route.The key to steady exit and accurate stop of unmanned buses lies in the planning and control of vehicle speed.In the urban structured road,how to plan and control the safe,comfortable and accurate stopping speed for buses is a problem that researchers from all walks of life pay great attention to.Therefore,this paper proposes a longitudinal motion planning and control algorithm for bus entrance and exit scenes.The main research contents are as follows:1)An urban structured road speed planning algorithm based on trapezoidal speed constraint is proposed.Firstly,the objective function of QP problem of speed planning is constructed based on S-T coordinate system,and the initial constraint conditions are used for preliminary optimization.Secondly,in order to solve the problems that the safety distance including actuator characteristics and body size is not considered in the constraint conditions,and the deceleration time before collision with obstacles is late,the dynamic obstacle trapezoidal speed planning algorithm based on safety distance is adopted to solve the speed.Finally,taking the solved speed as the nonlinear constraint of QP speed planning algorithm,the desired speed meeting the accuracy,safety and comfort of the inbound position is obtained.Simulation results show that the algorithm can effectively avoid static and dynamic obstacles in structured roads,and the planned parking position error is 0.13m.2)A joint learning algorithm of quality and slope parameters is proposed.Firstly,using the characteristic that the driving torque information of pure electric bus can be accurately obtained,the high-frequency analysis method is used to learn the vehicle quality parameters,thus decoupling the quality and slope.Secondly,the study of road slope parameters is realized by combining dynamic slope parameter learning based on state observation with kinematic slope parameter learning based on acceleration information.Finally,through the joint simulation of MATLAB/Simulink and Truck Sim,the test results show that the average absolute value error of quality learning is168.524kg,the average absolute percentage error is 2.56%,the average absolute value error of slope learning is 0.173°,and the average absolute percentage error is 2.08%,which meets the precision requirements of driving/braking switching strategy and model feedforward control.3)An active disturbance rejection longitudinal control algorithm based on parameter self-learning is proposed.Firstly,a control-oriented longitudinal dynamic model of pure electric bus is established,and a feedforward controller is designed based on the inverse model and the learning results of mass-gradient parameters to improve the responsiveness of longitudinal control.Secondly,the longitudinal system is highly nonlinear,dynamic response lag and uncertainty caused by aging variation.By designing ESO observer for disturbance compensation and designing feedback controller,the stability and adaptability of the control system are improved.Simulation results show that the designed control algorithm can reduce the parking position error by 85.2%.In each bus entrance and exit scene,the impact degree of the designed control algorithm is in the range of[-0.59m/s ~3,0.57m/s~3],it meets the requirements of driving comfort and stopping accuracy in different inbound and outbound scenarios.The results of real vehicle show that the control algorithm in this paper has good robustness and adaptability to low-speed straight-line driving,low-speed turntable driving,starting and leaving the station and stopping and entering the station. |
PDF Full Text Request