• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2026, Vol. 62 ›› Issue (14): 280-290.doi: 10.3901/JME.260483

• 运载工程 • 上一篇    下一篇

扫码分享

基于自适应动态规划的电液线控底盘博弈优化控制

周小川1, 储雨凯1, 赵万忠1, 王春燕1, 王敏1,2   

  1. 1. 南京航空航天大学能源与动力学院车辆工程系 南京 210016;
    2. 徐州徐工汽车制造有限公司 徐州 221112
  • 收稿日期:2025-09-01 修回日期:2026-01-02 发布日期:2026-08-29
  • 作者简介:周小川,男,1994年出生,博士,副研究员。主要研究方向为运载装备线控底盘、智能网联汽车规划控制。E-mail:zhouxiaochuan@nuaa.edu.cn;赵万忠(通信作者),男,1982年出生,博士,教授,博士研究生导师。主要研究方向为车辆动力学与控制、智能线控底盘。E-mail:zwz@nuaa.edu.cn
  • 基金资助:
    国家自然科学基金(U24A20199,52405112)和江苏省科技重大专项(BG2024038)资助项目。

Game Optimization Control of Vehicle Electro-hydraulic X-by-wire Chassis Based on Adaptive Dynamic Programming

ZHOU Xiaochuan1, CHU Yukai1, ZHAO Wanzhong1, WANG Chunyan1, WANG Min1,2   

  1. 1. Department of Vehicle Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016;
    2. Xuzhou XCMG Automobile Manufacturing Co., Ltd., Xuzhou 221112
  • Received:2025-09-01 Revised:2026-01-02 Published:2026-08-29

摘要: 线控底盘颠覆了传统汽车的机械式底盘构型,具有控制精度高、响应速度快、功能拓展多等显著优势,是高级辅助驾驶和智能驾驶的关键执行载体。针对底盘电动化、线控化水平低,协同控制能力差等问题,提出融合电液复合转向和电液复合制动的电液线控底盘构型,建立电液线控底盘直接横摆力矩与主动转向博弈优化框架,将车辆轨迹跟踪时考虑的多目标协同控制问题,转化为基于非合作纳什博弈的最优控制问题。针对线控底盘控制目标与过驱动异构执行器耦合下的最优控制求解难题,融合自适应动态规划和微分博弈理论,提出基于双启发式架构的线控底盘自适应动态规划博弈跟踪控制算法(Game tracking control based on adaptive dynamic programming,G-ADP),通过反向传播神经网络训练得到匹配最优控制器的执行网络,使线控底盘博弈系统性能指标达到纳什均衡。硬件在环试验结果表明,在不同附着系数的单移线和正弦参考轨迹测试工况下,所提的G-ADP算法能够减小跟踪误差并有效改善车辆稳定性。

关键词: 线控底盘, 博弈优化控制, 自适应动态规划, 神经网络, 轨迹跟踪

Abstract: The X-by-wire chassis has overturned the mechanical chassis configuration of traditional vehicles. It has significant advantages such as high control accuracy, fast response speed, and multiple functional expansion. It is a key execution carrier for advanced assisted driving and intelligent driving. In view of the problems of low electrification and wire control levels of vehicle chassis and poor collaborative control capabilities, a vehicle electro-hydraulic X-by-wire chassis configuration integrating electro-hydraulic composite steering and electro-hydraulic composite braking is proposed, and a direct yaw torque and active steering game optimization framework of the electro-hydraulic X-by-wire chassis is established. The multi-objective collaborative control problem in vehicle trajectory tracking is transformed into an optimal control problem based on non-cooperative Nash game. In view of the optimal control problem of the X-by-wire chassis under the coupling of control objectives and overdriven heterogeneous actuators, the adaptive dynamic programming and differential game theory are integrated to propose a game tracking control algorithm based on the dual heuristic architecture for the X-by-wire chassis adaptive dynamic programming game tracking control based on adaptive dynamic programming(G-ADP). The execution network matching the optimal controller is obtained through back-propagation neural network training, so that the performance indicators of the X-by-wire chassis game system reach Nash equilibrium. The hardware-in-the-loop test results show that the proposed G-ADP algorithm can reduce the tracking error and effectively improve the vehicle stability under the single-line change and sinusoidal reference trajectory test conditions with different adhesion coefficients.

Key words: X-by-wire chassis, game optimization control, adaptive dynamic programming, neural networks, trajectory tracking

中图分类号: