智能汽车非稳态纵向跟车行为滚动时域控制  

Receding horizon control of unsteady longitudinal following behavior for intelligent vehicle

作　　者：朱向雷[1] 王英资[1] 侯珏[1,2] ZHU Xianglei;WANG Yingzi;HOU Jue(China Automotive Technology&Research Center Co.,Ltd.,Tianjin 300300,China;School of Automotive Studies,Tongji University,Shanghai 201804,China)

机构地区：[1]中国汽车技术研究中心有限公司,天津300300 [2]同济大学汽车学院,上海201804

出　　处：《中国测试》2021年第5期118-122,共5页China Measurement & Test

基　　金：国家自然科学基金项目(51575229);中国汽车技术研究中心有限公司重点项目(16190125)。

摘　　要：为提升智能汽车非稳态纵向跟车性能并避免与前车发生碰撞,研究一种针对非稳态跟车行为的滚动时域控制方法。建立考虑动态延迟的自车运动学模型,并以车距误差、车速误差和自车加速度组成的误差向量为状态变量设计自车与前车纵向运动学关系模型。分析智能汽车非稳态跟车过程,并以解析形式给出非稳态跟车的可行性初始条件。基于跟车误差向量构建包含稳态车距、避免碰撞和加速度限值的非稳态跟车行为最优控制问题,并将其推导为包含误差向量初值参数、二次型目标函数、不等式约束和等式约束的二次规划问题,然后通过一系列有限时域内带约束的滚动优化问题得到非稳态跟车行为状态反馈控制律。结果表明,所设计的非稳态跟车控制方法能够有效控制自车速度和加速度,快速收敛至指定车距实现稳态跟车。In order to improve the performance of unsteady longitudinal car-following of intelligent vehicle and avoid collision with the front vehicle,a receding horizon control method for the unsteady following behavior was studied.The kinematic model of the self-propelled vehicle considering the dynamic delay was established,and the longitudinal kinematic relationship model of the self-propelled vehicle and the front vehicle was designed with the error vector composed of the distance error,the speed error and the acceleration of the selfpropelled vehicle as the state variable.The unsteady longitudinal car-following process of intelligent vehicle was analyzed,and the feasible initial conditions was given in analytical form.Based on the following error vector,the optimal control problem of the unsteady following behavior including the steady-state distance,collision avoidance and acceleration limit was constructed,and it was derived as a quadratic programming problem with initial parameters of error vectors,quadratic objective functions,inequality constraints and equality constraints.The unsteady following behavior with state feedback control law was obtained through a series of constrained receding optimization problems in the finite time domain.The results show that the designed control method can effectively control the speed and acceleration of the car and quickly converge to the specified distance to achieve the steady-state car following.

关 键 词：智能汽车 非稳态跟车 滚动时域控制 优化求解 

分 类 号：U27[机械工程—车辆工程]