【作者】 张杰；

【导师】 郁明；

【作者基本信息】 合肥工业大学 ， 控制理论与控制工程， 2021， 硕士

【摘要】 随着汽车工业和电子信息技术的高速发展,线控转向系统已经逐步走进大众的视野中。正是因为线控转向系统改变了传统转向系统的机械结构仅用电信号进行转向控制,使其越来越适用于智能汽车。车辆在转向过程中横摆稳定性控制是汽车控制的一个重要环节,线控转向汽车的主动前轮转向控制为改善车辆横摆稳定性提供了一个有效的方案。本文在滑模控制理论和极限学习机的基础上,提出了一种新型的线控转向汽车主动转向横摆稳定性控制策略来提高车辆的横摆稳定性和驾驶舒适性。本文首先介绍线控转向系统的工作原理及机械结构,对车辆在转向过程中进行动力学分析,考虑系统内在参数摄动和外部扰动,建立车辆动力学模型和线控转向系统的数学模型。接着,针对线控转向汽车的主动转向横摆稳定性控制设计了上下两层控制策略:上层设计了传统滑模控制器和自适应递归积分终端滑模控制器,同时控制车辆的侧滑角和横摆角速度来保证车辆的横摆稳定;下层构建了基于极限学习机的快速非奇异终端滑模控制器,作为转角跟踪控制器来确保实际前轮转角可以跟踪上上层控制器中计算得到的最佳前轮转角。上层的传统滑模控制器和自适应递归积分终端滑模控制器都是针对车辆的侧滑角和横摆角速度进行跟踪控制来保证车辆的横摆稳定性,都有着较好的控制效果。与传统的滑模控制器相比,自适应递归积分终端滑模控制器具有如下优势:1)在该滑模控制器中引入了积分项,不仅减少系统状态到达滑模面的时间,而且在一定程度上可以抑制抖振现象;2)依据递归思想设计的终端滑模面不仅保证了系统状态有限时间收敛性,而且同时控制了侧滑角和横摆角速度这两个对车辆的横摆运动有很大影响的系统状态,使得车辆的稳定性进一步提高;3)引入了自适应律,从李亚普诺夫意义上估计切换项增益,避免了传统滑模控制器设计中对干扰上界信息的要求,增加系统鲁棒性。下层使用基于极限学习机的快速非奇异终端滑模控制器来保证线控转向系统的实际前轮转角可以快速精确地跟踪上层控制器中获得的最佳前轮转角,其主要优势有:1)设计了快速非奇异终端滑模面,不仅保证了跟踪误差有限时间收敛,且收敛速度更快,也避免了奇异现象;2)控制策略的等效控制部分由极限学习机在线学习得到,大大减轻了在设计滑模控制器过程中对系统模型精确度的依赖性,而且可以保证控制器的性能。最后,根据实际的驾驶工况,设计四种仿真方案来模拟真实转向行为以及两种对比控制策略,使用MATLAB/Simulink和Car Sim软件进行联合仿真验证了本文所提出控制策略的有效性。从仿真结果可以看出,相比于传统的滑模控制器和仅考虑车辆横摆角速度跟踪情况的控制策略,本文所提出的新型控制方法可以提高线控汽车在转向过程中横摆稳定性,而且即使在出现外部强干扰时,车辆的横摆稳定性、驾驶舒适性仍能够得到保障。更多还原

【Abstract】 With the rapid development of vehicle industry and electronic information technology,the Steer-by-Wire(SBW)system has gradually come into the public’s view.Many intelligent vehicles tend to employ SBW system for its control flexibility since the mechanical structure of traditional steering system is replaced by electrical signals in SBW system.Considering the yaw stability of vehicles is also an important topic of the vehicle research,active front steering(AFS)control strategy for SBW vehicles has shown its reliability in improving vehicle stability in recent years.Thus,in this dissertation,a novel AFS control strategy based on sliding mode control(SMC)theory and extreme learning machine(ELM)is further studied for the purpose of improving the yaw stability and maneuverability of SBW vehicles.In this dissertation,the working principle and mechanical structure of the SBW system are firstly introduced.Then,the vehicle dynamics and mathematical model of SBW system are established by considering the system perturbation and external disturbances.Next,a conventional sliding mode control and adaptive recursive integral terminal sliding mode control(ARITSMC)are adopted in the upper controller for guaranteeing the convergence performance of both the actual sideslip angle and the yaw rate with strong robustness and fast convergence rate.And a fast nonsingular terminal sliding mode control(FNTSMC)with extreme learning machine(ELM)estimator to estimate its equivalent control is designed in the lower controller to track the desired front wheel steering angle calculated from the upper controller for driving the sideslip angle and the yaw rate to converge ideal value.The two upper controllers are designed to track the vehicle sideslip angle and yaw rate to ensure the yaw stability of SBW vehicles.Compared with the conventional sliding mode controller,the ARITSMC has better performance and its features are as follows: 1)Due to the introduction of an integral element in the proposed ARITSMC,not only can the reaching time be reduced,but also the chattering phenomenon can be decreased further.2)The tracking error can converge to zero in a finite time by designing the terminal sliding surface,and by combining the sideslip angle and yaw rate as the control objective,better stability performance can be obtained when compared with the case of using only one of these values alone.3)The switch gain in the upper controller can be estimated successfully in the sense of Lyapunov by using the adaptive law which relaxes prior knowledge of the system uncertainty required for conventional sliding mode controller design and enhances the system robustness.The lower FNTSMC based on ELM is implemented to guarantee the actual front wheel steering angle can track the desired one calculated from the upper controller.The features of this control scheme are as follows: 1)Compared the conventional sliding mode control,the proposed FNTSMC can guarantee the faster convergence rate and the error dynamics converge to the equilibrium points in a finite time rather than asymptotically.2)A novel estimator via the ELM is proposed to adaptively estimate the equivalent control component such that the dependence of system dynamics can be nicely alleviated.Finally,four simulation cases are designed to simulate the real steering behaviour according to the actual driving conditions,and two comparison control strategies are introduced to compare the performance of proposed controller by utilizing MATLAB/Simulink and Car Sim software.From the simulation results,it can be seen that the proposed control strategy obtains the most excellent performance compared the two other controllers.Furthermore,the yaw stability and maneuverability for SBW vehicles can be well maintained with the proposed controller when encountering strong external disturbances.更多还原