论文部分内容阅读
针对车辆自主驾驶中的车道线保持问题,基于车辆横向动力学简化线性模型,考虑前向轮自动转动系统的惯性延迟特性,首先将模型按照实际情况分为相对确定与不确定两个子系统,提出一类反演控制与滑模控制相融合的横向位置误差复合控制方法。该方法能保留滑模控制对不确定横向动力学模型的强鲁棒性优点;同时,针对确定的惯性延迟系统,又能继承反演控制层层倒推、环环相扣、精确巧妙设计的思想。由于柔化函数的引入,减弱了传统滑模控制的颤振问题。通过构造Lyapunov函数,保证了系统所有闭环信号均有界。最后,针对轮胎刚度系数不确定的情况,进行了100次随机数字仿真,结果验证了所提方法的良好鲁棒性与快速性。值得说明的是,把复杂系统按照物理实际情况分解为相对确定与不确定两部分的思想,对自动驾驶中车道线保持问题非常有物理意义,能够为未来的车辆试验提供理论参考。
Aiming at the problem of lane keeping in vehicle autonomous driving, based on the simplified linear model of vehicle lateral dynamics and considering the inertial delay characteristics of the automatic rotating system of front wheel, the model is divided into two subsystems of relative determination and uncertainty according to the actual situation, A Hybrid Method of Lateral Position Error Control Based on Inversion Control and Sliding Mode Control. This method can preserve the strong robustness of sliding mode control to uncertain lateral dynamics model. In the meantime, for the established inertial delay system, it can inherit the inversion control layer backstepping, interlocking and precise ingenious design thought. Due to the introduction of softening function, the traditional sliding mode control flutter problem is weakened. By constructing Lyapunov function, we ensure that all the closed-loop signals of the system are bounded. Finally, 100 times of random number simulation are carried out for the case of tire stiffness coefficient uncertainty. The results verify the robustness and rapidity of the proposed method. It is noteworthy that the idea of decomposing a complex system into two parts: relative determination and uncertainty depends on the actual situation of the physical system. It is of great physical significance to the lane keeping problem in automatic driving and can provide a theoretical reference for future vehicle tests.