Remarks on the Theory of Vehicle Vibrational Analysis Based on the On-the-Road Measurements

SUMMARY

Basic relation between input spectral density matrix and output spectral density matrix of a linear stochastically excited dynamic system is indicated. General conclusions regarding the output processes spectral densities, coherences and phase angles in respect to the input processes stochastic properties are drawn. The possibility of the determination of the system's transfer functions when input and output spectral density matrices are known is discussed. Applications of the obtained results in vehicle vibrational analysis when the vehicle is considered as one-input, two-input or multiinput system are shown.     

Power Spectral Density of Road Profiles

The road profile is usually considered to be a random process x ( d ), where x is the road height and d is the distance along the road. As the vehicle travels along the road with velocity v , the random process x ( d ) is converted to a random process x ( t ) which is input to the vehicle suspension via the tyre. The random process x ( d ) is usually described in terms of its power spectral density as a function of frequency in either radians or cycles per unit distance. However, there are several different ways of defining power spectral density, and this makes it difficult to compare published data without knowing how the power spectral density has been defined. The proper calculation of RMS values of vehicle response for an assumed road power spectral density is explained by an example.     

Power Spectral Density of Road Profiles

The road profile is usually considered to be a random process x (d), where x is the road height and d is the distance along the road. As the vehicle travels along the road with velocity v, the random process x (d) is converted to a random process x (t) which is input to the vehicle suspension via the tyre. The random process x (d) is usually described in terms of its power spectral density as a function of frequency in either radians or cycles per unit distance. However, there are several different ways of defining power spectral density, and this makes it difficult to compare published data without knowing how the power spectral density has been defined. The proper calculation of RMS values of vehicle response for an assumed road power spectral density is explained by an example.     

Optimal Preview Control of a Two-dof Vehicle Model Using Stochastic Optimal Control Theory

An optimal preview control algorithm is applied to a two degree of freedom(dof) vehicle model travelling with constant velocity on a randomly profiled road. The road roughness is modelled as a homogeneous random process being the output of a linear first order filter to white noise. The input from the road irregularity is assumed to be measured at some distance in front of the vehicle and this measured infonnation is utilized by the active controller to prepare the system for the ensuing input. The preview control algorithm is obtained by minimizing a quadratic performance index and by describing the average behaviour of the system by the covariance matrix of the vehicle response state vector. Results are presented for full state feedback and significant improvements in sprung mass acceleration, suspension working space and road holding are observed.     

基于白适应神经模糊推理系统的跟驰模型研究

由于驾驶行为的不确定性,难以建立精确的车辆跟驰模型。针对这一问题,应用自适应模糊神经推理系统（ANFIS）建立跟驰模型,以跟随车与前车速度差及行车间距为输入量、跟随车的加速度为输出量,建立25条模糊推理规则,将模糊推理规则产生的数据作为车辆跟驰ANFIS模型的训练数据,并利用MATLAB编程对其进行训练。最后,设计了基于车载高精度GPS的跟驰试验,并结合试验数据分别对自适应模糊神经推理系统跟驰模型和传统跟驰模型进行仿真。结果表明,前者输出的跟驰车辆加速度值更接近于真实值。     

基于小波变换的四轮车辆非平稳振动时频研究

基于8自由度的车辆动力学模型，以四轮相关滤波白噪声路面为系统输入，对四轮车辆的非平稳随机振动过程进行了计算机时域模拟，进而采用谱分析和小波分析技术分析了系统响应，在时频平面内建立了各振级之间的振动联系，考察了振动能量的流动情况，深入揭示了车辆非平稳振动过程的本质。     

基于自适应神经模糊推理系统的跟驰模型研究

由于驾驶行为的不确定性,难以建立精确的车辆跟驰模型.针对这一问题,应用自适应模糊神经推理系统(ANFIS)建立跟驰模型,以跟随车与前车速度差及行车间距为输入量、跟随车的加速度为输出量,建立25条模糊推理规则,将模糊推理规则产生的数据作为车辆跟驰ANFIS模型的训练数据,并利用MATLAB编程对其进行训练.最后,设计了基于车载高精度GPS的跟驰试验,并结合试验数据分别对自适应模糊神经推理系统跟驰模型和传统跟驰模型进行仿真.结果表明,前者输出的跟驰车辆加速度值更接近于真实值.     

基于汽车车身垂直加速度的典型道路路面谱识别研究

以1/4汽车振动模型为研究对象,推导出以汽车车身垂直振动加速度作为输入信号、路面不平度作为输出信号的数学模型及其模拟图,并利用MATLAB/SIMULINK搭建系统模型求解路面不平度,对路面不平度进行谱估计完成路谱的识别。通过实际测试和数据处理分析,说明该方法理论依据正确可行,可以为虚拟样机仿真路面的生成提供数据支持。     

某车型散热器框架振动疲劳分析

运用CAE虚拟仿真技术,建立精确的FEA模型,利用采集整车道路试验加速度功率谱密度信号对汽车散热器框架进行振动疲劳分析,数值仿真与整车道路试验结果表明虚拟疲劳寿命仿真结果可靠性可作为设计优化的依据。     

On Inverse Equations for Vehicle Dynamics

Instead of writing equations which when solved yield the response of a vehicle to an input such as the front wheel steer angle, one can often invert the equations so that a response quantity is specified as an input and a new set of equations is solved yielding the steer angle required as an output. Using these equations one can discover the input steer angle a driver would need to impose in order to accomplish a specific maneuver for various vehicles.

It is shown that there are many possible inverse equation sets and that the eigenvalues of the inverse equations are hard to interpret since they may have little to do with the vehicle parameters. The linear single-input single-output case is studied first to fix ideas using a simple example. For the bicycle model vehicle, it is shown that any vehicle may have unstable inverse equations depending upon the response quantity used. Extensions to nonlinear and multiple-input multiple output systems are discussed.     

On Inverse Equations for Vehicle Dynamics

Abstract

Instead of writing equations which when solved yield the response of a vehicle to an input such as the front wheel steer angle, one can often invert the equations so that a response quantity is specified as an input and a new set of equations is solved yielding the steer angle required as an output. Using these equations one can discover the input steer angle a driver would need to impose in order to accomplish a specific maneuver for various vehicles.

It is shown that there are many possible inverse equation sets and that the eigenvalues of the inverse equations are hard to interpret since they may have little to do with the vehicle parameters. The linear single-input single-output case is studied first to fix ideas using a simple example. For the bicycle model vehicle, it is shown that any vehicle may have unstable inverse equations depending upon the response quantity used. Extensions to nonlinear and multiple-input multiple output systems are discussed.     

Small Vibrations of Nonlinear Dynamic Systems Containing Non-Rigid Bodies Excited by Stationary Stochastic Processes

From a vibrational point of view, real vehicles form dynamic systems containing non-rigid bodies and nonlinear two-force elements, often excited by stationary random processes. Their theoretical investigation in a broader frequency range is practically possible in the frequency domain only. In the treatment of nonlinearities second-order stochastical linearization methods can be used advantageously. The amount of necessary computations can be vastly reduced by using known receptance matrices of all bodies in the system in its equations of motion after Fourier transformation. The article describes the receptance properties of non-rigid bodies, their use for the determination of the dynamic system's transfer functions and the determination of the output processes' spectral density matrix under stationary random external excitation using stochastical second-order linearization of the nonlinearities.     

轿车悬架阻尼二阶最优控制设计方法研究

本文以半动悬架与二阶最优控制理论为基础，通过被动悬架阻尼的主动力表示，结合对最优控制力的分解以及最优评价指标，振动加速功率谱密函数与系统振动阻尼之间的关系的研究，建立了多自由度汽车振动系统在随机输入下以振动加速度功率谱密度函数极小为目标的悬架最优减振器阻尼参数的方法，并给出了７自由汽车振动最优减振器阻尼设计实例。     

基于DSP的车辆噪声信号分析检测研究

从车辆噪声源入手,分析了车辆噪声信号的差异性和声传感器采集的可行性,采用TMS320F2812数字处理芯片辅以外部电路,通过内置模数转换模块采集外部车辆噪声信号,并将其在计算机终端进行预处理、功率谱密度分析以及小波特征提取算法仿真,最终将验证后的算法程序下载到TMS320F2812的FLASH中独立运行。将此系统用于车辆类型判断、车辆结构损坏检测、车辆超载或者超速监视等,取得了满意的监测效果。     

驾驶机器人车辆动态制动力矩补偿

为了减小长期自动驾驶过程中制动性能下降带来的影响,提出了一种驾驶机器人车辆动态制动力矩补偿方法。首先建立了以车速和制动踏板力为输入,制动力矩为输出的驾驶机器人车辆制动性能离线自学习模型。然后考虑到驾驶机器人车辆长期自动驾驶导致离线自学习模型可靠性下降,建立了以车速和制动踏板力为输入,制动力矩为输出的扩展自回归在线辨识模型,并采用模糊变遗忘因子递推最小二乘法进行参数辨识。模糊变遗忘因子递推最小二乘法通过引入遗忘因子的方式,对数据施加时变加权系数,以避免出现数据增长导致的数据饱和现象。模糊变遗忘因子控制器以制动力矩辨识误差为输入,经模糊规则推理实时输出合适的遗忘因子进行参数辨识,能够有效均衡驾驶机器人车辆制动性能参数辨识的稳定性与收敛速度。驾驶机器人车辆自动驾驶过程中,根据当前车速与目标车速的大小计算出所需的制动力矩,加上反馈回来的制动力矩误差,并结合当前时刻的车速,利用制动性能离线自学习模型与机械腿逆向运动学模型实时计算出制动电机输出位移量,实现对驾驶机器人车辆制动力矩的在线补偿。仿真与试验结果表明:利用所提出的方法对车辆动态制动力矩进行辨识时,通过调节遗忘因子,辨识结果能够快速收敛且辨识误差较小。在此基础上,控制驾驶机器人车辆进行纵向车速跟踪时,能够有效减小制动性能下降造成的影响,保证控制车速跟踪误差在±1km·h-1之内。     

Mazda6轿车转向传感器信号的simulink识别

转向传感器信号是车辆控制系统的重要输入信号,但转向传感器的输出不是直接给出转向角大小而是方波信号。为了探究转向传感器原理以及Mazda6轿车实际转向过程中转向传感器信号,本文将介绍如何用simulink来识别该方波信号,输出转向角,最后用该方法识别出实际转向过程的转向角大小。     

Spectral Decomposition Methods for the Computation of RMS Values in an Active Suspension

The theory of the spectral decomposition of a matrix with distinct eigenvalues is reviewed briefly and then applied to the closed-loop system matrix of a quarter-car active suspension with preview. RMS values of the state variables and the control force are derived for an integrated white noise type random road input with the assumption of an optimised control system. The explicit formulæ obtained are easily computed numerically using MATLAB with accurate results at all speeds.     

Analysis of Tire Effect on the Simulation of Vehicle Straight Line Motion

In the dynamic simulation of vehicle straight line motion, a vehicle model usually drifts from its intended straight path even in the case of no external input. This is particularly true when a tire model based on experimental data is used. The purpose of this paper is to provide an enhancement of a basic understanding of a tire/vehicle system behavior in the straight line motion and to identify the effect of the tire on that motion. Through the analysis of a two degrees of freedom vehicle model, tire characteristic which causes a lateral drift in the straight line motion is identified. Then the results are confirmed from vehicle test and the simulations with a more complex full-car model.     

Analysis of Tire Effect on the Simulation of Vehicle Straight Line Motion

In the dynamic simulation of vehicle straight line motion, a vehicle model usually drifts from its intended straight path even in the case of no external input. This is particularly true when a tire model based on experimental data is used. The purpose of this paper is to provide an enhancement of a basic understanding of a tire/vehicle system behavior in the straight line motion and to identify the effect of the tire on that motion. Through the analysis of a two degrees of freedom vehicle model, tire characteristic which causes a lateral drift in the straight line motion is identified. Then the results are confirmed from vehicle test and the simulations with a more complex full-car model.     

基于RBF神经网络识别路面谱的新方法

路面不平度是车辆行驶中振动的重要激励。为了识别路面不平度的功率谱密度函数(路面谱),提出了一种基于径向基函数(RBF)神经网络识别路面谱的新方法。该方法以7自由度汽车振动模型为基础,以MATLAB软件仿真得到的汽车车身质心垂直加速度谱为神经网络理想输入样本,以GB7031-86建议的路面谱为神经网络理想输出样本,应用RBF神经网络建立汽车车身质心垂直加速度谱和路面谱之间的非线性映射模型。另取一组仿真得到的车身质心垂直加速度谱代入已训练好的网络进行路面谱识别。结果表明:该方法具有较强的抗噪声能力和较理想的识别精度,识别的路面谱与拟合的路面谱吻合一致。