摘要
针对自动驾驶汽车在道路大曲率弯道下跟踪精度不高的问题,聚焦于道路曲率对横向控制策略的影响,分别从车辆模型建模、横摆稳定性和时域优化三个角度对基于传统模型预测控制(MPC)算法的横向控制策略进行了改进优化。将道路曲率融入车辆模型,建立了曲率前馈的误差动力学模型,并以此为基础设计了基于曲率前馈MPC算法的横向控制策略。然后在策略中添加了一个由横向车速和稳态横摆角速度组成的横摆稳定性约束来提高车辆在大曲率工况时的横摆稳定性。基于遗传算法建立了车速、道路曲率和时域三者之间的MAP图,以优化策略的预测时域和控制时域。进行了仿真分析,结果表明改进后的横向控制策略能够有效提高车辆的路径跟踪精度和横摆稳定性。最后,实车道路试验验证了曲率前馈MPC策略的有效性。
Aiming at the problems of low tracking accuracy of autonomous vehicles in the road with large curvature curves,the influences of road curvature on the lateral control strategy were focused,the lateral control strategy was improved and optimized based on the traditional model predictive control(MPC)algorithm from three aspects of vehicle model modeling,yaw stability and time domain optimization,respectively.The road curvature was integrated into the vehicle model,and an error dynamics model with curvature feedforward was established.And then,a lateral control strategy was designed based on curvature feedforward MPC algorithm.Then,a lateral stability constraint consisting of lateral vehicle speed and steady-state lateral angular velocity was added to the strategy to enhance the lateral stability of the vehicles under high curvature conditions.A MAP map was established based on genetic algorithm to optimize the prediction and control time domains of the strategy,taking into account the relationships among vehicle speed,road curvature and time domain.Simulation analysis was conducted,and the results show that the improved lateral control strategy may effectively improve the path tracking precision and lateral stability of the vehicles.Finally,the effectivenesses of the curvature feedforward MPC strategy were verified through real vehicle road tests.
作者
林歆悠
金忠伟
唐云亮
LIN Xinyou;JIN Zhongwei;TANG Yunliang(School of Mechanical Engineering and Automation,Fuzhou University,Fuzhou,350108)
出处
《中国机械工程》
2025年第11期2774-2782,2791,共10页
China Mechanical Engineering
基金
国家自然科学基金(52272389
51505086)
福建省自然科学基金(2020J01449)
载运工具与装备教育部重点实验室开放课题(KLCE2022-08)
四川省新能源汽车智能控制与仿真测试技术工程研究中心开放课题(XNYQ2022-001)
安徽工程大学检测技术与节能装置安徽省重点实验室开放研究基金(JCKJ2021A04)。
关键词
曲率前馈
横向控制策略
车辆模型建模
横摆稳定性
时域优化
curvature feedforward
lateral control strategy
vehicle model modeling
yaw stability
time domain optimization
作者简介
通信作者:林歆悠,男,1981年生,教授、博士研究生导师。研究方向为新能源汽车电驱动控制策略、智能驾驶轨迹追踪与转向决策控制、融合燃料电池衰退和动态特性的能量管理策略。E-mail:linxinyoou@fzu.edu.cn。
