【作者】 尹浩；

【导师】 温卫东；

【作者基本信息】 南京航空航天大学 ， 车辆工程， 2007， 博士

【摘要】 针对汽车操纵逆动力学的研究现状,采用仿真分析和实车试验相结合的方法,对汽车操纵逆动力学进行了比较系统的研究。将计算智能方法、控制论中的逆系统理论以及虚拟样机技术应用于汽车操纵逆问题的研究中。在操纵逆问题研究基础上,进行基于逆问题求解的汽车操纵性能的优化,并根据优化方案,分析汽车运动稳定性。主要的研究内容和成果有:1.以识别汽车方向盘转角输入和方向盘转矩输入为操纵逆动力学研究的出发点。通过求解人—车—路闭环系统的状态空间表达式,利用径向基函数网络建立了汽车横摆角速度、侧向加速度与方向盘转角之间的映射关系。在汽车沿不同路径行驶时,以汽车横摆角速度、侧向加速度为输入,识别方向盘转角,结果表明,这种求逆的方法是可行的,并且具有精度高、运算速度快及抗干扰能力强等优点。2.根据车辆的操纵稳定性能以及汽车的结构特点,利用ADAMS软件建立了包含悬架系统、转向系统以及轮胎等结构在内的整车模型;根据ADAMS闭环仿真控制原理,以汽车行驶道路轨迹为输入控制,识别方向盘转角,结果表明,这种方法是可行的,为汽车操纵逆动力学研究提供了可以信任的整车模型。3.以二自由度汽车为研究对象,将控制论中的逆系统理论应用于操纵逆问题的研究。将汽车操纵逆问题转化为建立原系统的逆系统,并在该逆系统中求解输出,即归结为正问题的处理。在建立的逆系统中,以侧向加速度为输入,求解汽车方向盘转角,结果表明,该方法是有效的,且精度比较高,运算速度比较快。4.将上述三种方法识别的方向盘转角进行对比分析,并通过实车试验数据验证。可以看出,三种方法识别的方向盘转角比较吻合,且数字仿真结果与实车试验数据具有较好的一致性,数字仿真能较好的反映实车试验。5.运用计算智能方法,以实车实验为依据,进行汽车操纵逆动力学的研究,建立了基于试验数据的径向基函数网络。由横摆角速度、侧向加速度的试验数据,识别汽车方向盘转角,将识别结果与试验测得的方向盘转角相对比,对比结果验证了该方法的正确性,并且只要试验样本足够多且具有代表性,识别精度会越来越高。6.给出了一种基于逆问题求解的闭环系统操纵性能优化的方法。由跟踪路径反求出方向盘转角及汽车的其它响应,进而计算闭环系统的操纵性能评价指标并进行优化。该方法是在不同汽车方案具有相同实际行驶路径的基础上对操纵性能进行分析并优化,从而得到的最优汽车方案在跟踪某一典型路径时具有最好的操纵性能。在此基础上,根据优化得到的汽车方案,在汽车转向运动力矩输入模型和转角输入模型基础上,对汽车在不同附着系数路面上的运动稳定性进行分析。更多还原

【Abstract】 A brief review of the developing history of vehicle handling inverse dynamics is introduced in this paper. Using computer simulation technology and experimentation design method, the inversion solution study of vehicle handling dynamics is carried out, based on the method of computation intelligent algorithms, inverse system theory and virtual prototype technology. After the solution of inverse problems, an optimization approach is proposed for the purpose of improving the vehicle maneuverability, and then the motion stability of vehicle driving is analyzed based on the result of optimization.The main contributions of this work are summarized as follows:1. A method of obtaining the steering wheel angle input and steering moment input is presented for further investigation of the vehicle handling inverse dynamics, according to some vehicle responses. Using Radial Basis Function neural networks, the mapping relationship between yaw velocity, lateral acceleration and steering wheel angle is founded. The inverse solution results showed that the proposed inverse solution method is not only practical, but also with high accuracy, little computation requirement and good stability.2. Based on the characteristic of handling-stability performance and the vehicle configurations, the parameter 3D model of the vehicle is created successfully, which includes suspension, steering, tire etc. According to the mechanism of ADAMS closed-loop control, the steering angles can be identified with the input of road track. The identification shows this method is effectual, and the trustful vehicle model for handling inverse problems is given.3. Based on the two-degree-freedom system of vehicle, the inverse system theory is applied for the vehicle handling inverse dynamics. An inverse system is founded relativing to the primary system and the output of the inverse system is solved. The relationship between lateral acceleration and steering angle can be found in the inverse system, and the inverse solution results shows that this method is not only applicable, but also with high accuracy, little computation requirement.4. The identification results of the three methods, which include computation intelligent algorithms, inverse system theory and virtual prototype technology, is compared with each other and validated by the real experiment. The comparison shows that the steering angles which are solved through the three methods are approximate. The value of simulation is consistent with real experiment.5. Using the method of computation intelligent algorithms, the inversion solution study for the real experiment is carried out. The Radial Basis Function neural networks are established based on the test of vehicle handling and stability. The data of yaw velocity, lateral acceleration are put into the networks, and the identification shows its validity. The precision of identification can be higher, if the experimental data are enough and representative.6. Based on solution of inverse problems, an optimization approach is proposed for the purpose of improving maneuverability of driver-vehicle-road closed-loop systems. The mapping relationship between vehicle lateral displacement and steering wheel angle and other responses can be found utilizing Radial Basis Function neural networks. One prescribed path is taken as input of the trained RBF neural networks, then the steering wheel angle and other vehicle responses can be obtained and the maneuverability index of the closed-loop system can be obtained and optimized. It can be seen that different vehicle configurations, based on the inversion solution study, have the inherent ability to follow the same prescribed path, therefore the optimal vehicle configuration has the best maneuverability among all vehicle configurations when they follow some typical path. After the optimization, the motion stability of vehicle driving on different roads is analyzed based on the modal of vehicle with steering angle input and steering moment input.更多还原