J Shanghai Jiaotong Univ Sci ›› 2023, Vol. 28 ›› Issue (1): 114-125.doi: 10.1007/s12204-023-2574-2

• • 上一篇    下一篇

恶劣行驶条件下无人车辆路径跟踪串级优化控制

  

  1. (1. 广西科技大学 电气与信息工程学院;广西汽车零部件与整车技术重点实验室,广西 柳州 545006;2. 华南理工大学 机械与汽车工程学院,广州 510641)
  • 收稿日期:2022-03-21 出版日期:2023-01-28 发布日期:2023-02-10

Cascade Optimization Control of Unmanned Vehicle Path Tracking under Harsh Driving Conditions

HUANG Yinggang1 (黄迎港), LUO Wenguang1∗ (罗文广), HUANG Dan2 (黄 丹), LAN Hongli1 (蓝红莉)   

  1. (1. School of Electrical and Information Engineering; Guangxi Key Laboratory of Auto Parts and Vehicle Technology, Guangxi University of Science and Technology, Liuzhou 545006, Guangxi, China; 2. School of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou 510641, China)
  • Received:2022-03-21 Online:2023-01-28 Published:2023-02-10

摘要: 在超高速以及恶劣行驶条件下,传统控制方法难以同时保证无人车在转弯过程中的路径跟踪精度和驾驶稳定性。因此,本研究提出一种无人车辆路径跟踪串级控制策略。考虑基于车辆轮胎侧滑和道路曲率影响建立一种新的车辆误差模型,采用前馈-参数自适应线性二次调节器(LQR)和比例积分(PI)控制算法设计速度保持控制器来组成无人车路径跟踪串级控制器。为提高无人车在恶劣驾驶条件下的适应性,使用BP神经网络自动调整LQR控制器参数,其中初始权值和阈值根据驾驶条件使用改进的灰狼优化算法进行优化;保速控制器提高了在非线性车速变化下的曲线跟踪精度。最后,建立MATLAB/Simulink和CarSim的联合模型,仿真结果表明:所提出的控制方法可以实现无人车在超高速下的保速过弯能力。在强风和冰雪路面条件下,该方法表现出更高的跟踪精度,与前馈-LQR、单点预瞄控制和纯追踪控制等方法相比,在驾驶和变曲率路面上对外界干扰的适应性和鲁棒性更强。

关键词: 无人驾驶车辆,路径跟踪,恶劣行驶条件,串级控制,改进灰狼优化算法,BP神经网络

Abstract: Under ultra-high-speed and harsh conditions, conventional control methods struggle to ensure the path tracking accuracy and driving stability of unmanned vehicles during the turning process. Therefore, this study proposes a cascade control to solve this problem. Based on the new vehicle error model that considers vehicle tire sideslip and road curvature, the feedforward-parametric adaptive linear quadratic regulator (LQR) and proportional integral control-based speed-keeping controllers are used to compose the path-tracking cascade optimization controller for unmanned vehicles. To improve the adaptability of the unmanned vehicle path-tracking control under harsh driving conditions, the LQR controller parameters are automatically adjusted using a backpropagation neural network, in which the initial weights and thresholds are optimized using the improved grey wolf optimization algorithm according to the driving conditions. The speed-keeping controller reduces the impact on the curve-tracking accuracy under nonlinear vehicle speed variations. Finally, a joint model of MATLAB/Simulink and CarSim was established, and simulations show that the proposed control method can achieve stable entry and exit curves at ultra-high speeds for unmanned vehicles. Under strong wind and ice road conditions, the method exhibits a higher tracking accuracy and is more adaptive and robust to external interference in driving and variable curvature roads than methods such as the feedforward-LQR, preview and pure pursuit controls.

Key words: unmanned vehicles, path tracking, harsh driving conditions, cascade control, improved gray wolf optimization algorithm, backpropagation neural network

中图分类号: