Abstract: In response to the poor adaptability of conventional path tracking model predictive controllers under high-speed and large-curvature conditions, an adaptive prediction horizon control strategy was proposed based on optimized dynamic models. Firstly, to address the issues of insufficient accuracy of classical dynamics models under high lateral acceleration conditions, an optimized model including roll steer and compliance steer was established, achieving higher precision prediction of vehicle states. Secondly, to address the issues of fixed prediction horizon control under high-speed and large-curvature conditions, an adaptive prediction horizon strategy was proposed based on two-dimensional Gaussian function, achieving real-time adjustment of preview distances with low algorithm complexity. Finally, the effectiveness of the controller on double-lane-change roads was verified throught CarSim/Simulink joint simulation. Results show that a reduction of 45.1% in lateral position peak errors and 72.4% in yaw angle peak errors indicate better adaptability of the designed controller to extreme conditions.

Key words: intelligent connected vehicle, lateral dynamics optimization, path tracking, model predictive control, adaptive prediction horizon

摘要： 针对一般路径跟踪模型预测控制器在高速大曲率工况下适应性差的问题，提出了一种基于优化动力学模型的自适应预测时域控制策略。首先，为解决经典动力学模型在较高侧向加速度工况下精度不足的问题，建立了包含侧倾转向和变形转向特性的优化模型，实现了车辆状态的较高精度预测；其次，为解决高速大曲率工况下固定预测时域控制效果不佳的问题，提出了基于二维高斯函数的自适应预测时域策略，以低算法复杂度实现了预瞄距离的实时调整；最后，通过CarSim/Simulink联合仿真实验验证了控制器在双移线道路上的控制效果，结果表明，横向位置峰值误差降低45.1%，横摆角峰值误差降低72.4%，设计的控制器对极限工况有更好的适应性。

关键词: 智能网联汽车, 横向动力学优化, 路径跟踪, 模型预测控制, 自适应预测时域