基于轮边驱动电动汽车轮胎力估计的路面附着系数估算

为了能够实时准确的获得当前车轮的轮胎力及路面附着系数以提高汽车主动安全性能,提出一种轮边驱动电动汽车状态估计与路面附着系数估计相结合的估计方法。根据车载传感器及七自由度非线性车辆动力学模型,采用扩展卡尔曼滤波算法(EKF)进行车辆状态及轮胎力的估计。结合EKF估算结果和轮胎模型,采用递归最小二乘法(RLS)实时估计不同路面的附着系数。仿真结果表明:该方法可以在较为复杂工况下估计出不同的路面附着系数,估计精度较高,实时性较好。     

路面附着系数识别方法发展现状综述

实时、准确地识别路面附着系数是国内外汽车行业研究领域的热点和难点之一。针对目前国内外路面附着系数识别领域研究现状,根据汽车运动工况及动力学原理,分别对汽车纵向运动过程中的路面附着系数识别方法和汽车侧向运动过程中的路面附着系数识别方法进行了论述。对路面附着系数识别中常涉及的附着率估算、路面类型识别方法进行了分析,最后对路面附着系数识别技术发展方向进行了讨论。     

汽车ABS制动过程中道路识别方法研究

车辆防抱死控制系统（ABS）的目标控制参数在不同路面上存在很大差异,所以在不同路况下汽车电控系统所采取的控制策略和算法也有所差别。以汽车主动安全装置ABS为基础,在建立了车辆模型和进行滑移率估算的前提下．设计了道路识别控制器。考虑到轮胎非线性的影响,对变附着系数路面进行了ABS制动模拟试验。结果表明：基于路面识别技术的ABS控制系统能准确判断出路面状况,并据此调整控制策略,以使车辆获得最大的制动减速度和最短的制动距离。试验表明．该系统具有较好的跟踪性。     

基于轮胎回正力矩的路面附着系数估计方法

路面附着系数的识别对汽车稳定性控制起着至关重要的作用。轮胎回正力矩能够反映整车及轮胎的运动、受力状况以及路面环境信息,且利用回正力矩能比使用侧向力更早地估计轮-地接触状况。为此,本文设计一种基于轮胎回正力矩的路面附着系数估计方法。首先,基于二自由度车辆模型设计轮胎侧偏角反馈观测器,对轮胎侧偏角进行实时估计;其次,基于轮胎侧偏角和轮胎回正力矩信息设计路面附着系数估计器,构建路面附着系数评估函数;最后,搭建Carsim与Simulink联合仿真平台,仿真结果表明设计的估计算法能够有效地对路面附着系数进行估计。     

基于状态观测器的半挂汽车折叠角估算

根据商用车ESC系统各传感器的信息,利用半挂汽车3自由度模型设计了状态观测器来估计半挂汽车的折叠角.根据侧向加速度值对当前利用路面附着系数进行了估算,用来修正线性轮胎模型.利用三轴半挂汽车的21自由度非线性模型对观测器的估计结果进行了验证.结果表明,该观测器在车辆的线性和非线性区域内都能准确地估算出半挂汽车的折叠角.     

低附着路面电动助力转向系统助力控制研究

对传统的电动助力转向系统,当车辆在低附着系数路面上转向时,转向阻力矩会大幅降低,容易导致车辆侧滑甚至侧翻,因此助力控制必须考虑附着系数的影响。本文中在建立了整车动力学模型和Dugoff轮胎模型的基础上,利用横摆角速度和轮速实时估算出当前的附着系数,并据此设计了电流补偿助力模糊控制器。在MATLAB/Simulink环境下构建了系统的仿真模型并进行仿真。结果表明,在低附着系数路面上,该控制策略可在保证转向轻便性的前提下提高驾驶员的转向力矩,且附着系数越小或车速越高,转向力矩增加的程度越大,有效地防止了车辆转向过度,提高了路感。     

A Research on Road Condition Identification Based on Characterization Factors

在3种经典轮胎-路面数学模型的基础上,引入了一种“路面状态特征因子”的概念(它代表最佳滑移率前附着系数-滑移率曲线段下面的封闭面积),给出了7种典型路面的特征因子阈值及其区间,据此识别汽车当前行驶路面状态.基于8自由度汽车动力学模型,分别在单一路面和对接路面上进行制动模拟试验.结果表明该方法能较准确快速地识别路面状态.最后在自制的汽车防抱制动装置试验台上进行的制动试验进一步验证了方法的有效性和可行性.     

路面附着性能影响因素分析及其改善对策的研究

路面附着性能是决定汽车安全行驶的必要条件。影响轮胎与路面间附着性能的因素很多,主要包括轮胎、路面及汽车行驶状态等等。针对上述影响因素,分别对路面类型、路面状况、轮胎结构、胎面花纹及花纹深度、轮胎气压和车速等主要影响因素进行了深入分析,研究了各因素之间的内在联系和变化特点,揭示了各因素对路面附着性能的影响规律,并从路面和轮胎两个方面提出了改善轮胎与路面间附着性能的对策,从而提高汽车行驶的安全性。     

汽车双转鼓惯性试验台等效路面附着机理研究

本文中针对一种双转鼓惯性试验台,基于CAN/PCI总线建立机电一体化分布式测控系统,用于车辆安全性、动力性试验检测。基于所设计惯性试验台提出通过改变车轮与双转鼓间安置角等效不同路面峰值附着系数的算法,基于制动过程单轮动力学分析,建立单轮-试验台系统动力学模型,根据动力学模型获得可变安置角与路面峰值附着系数等效路面附着机理,基于Matlab/Simulink建立单轮-试验台系统仿真模型,验证了等效路面附着系数算法。通过所建单轮-试验台系统进行了试验,在小滑动区实时获取车轮所受纵向力及滑动率,基于Slip-slope理论实时估算等效峰值附着系数。结果表明,双转鼓惯性试验台等效路面附着机理正确,等效路面附着系数算法准确可行。     

基于智能轮胎系统的实时路面辨识技术

在复杂和极限工况下,路面附着系数是进行轮胎受力分析和车辆动力学控制的重要状态参数。相对于模型估计的方法,智能轮胎技术能够将轮胎与路面的交互信息反馈给车辆控制系统。本文提出了一种将智能轮胎系统和机器学习相结合的车辆路面附着系数获取方法。首先,考虑行驶工况环境进行传感器选型,开发基于MEMS三轴加速度传感器的智能轮胎硬件采集系统,并采用简化硬件结构的无线传输模式。其次,通过采集不同路面上的实车实验数据进行车辆实验收集机器学习训练的数据集,并分析轮地关系及信号特征。最后,将CNN与LSTM两者的优势相结合实现了对加速度时序信号的特征学习。通过与其它神经网络模型训练结果的比较,验证了所提CNN-LSTM双通道融合神经网络模型的有效性和准确性。本文提出的路面辨识方案实现了实时道路识别的目标,硬件与软件架构和神经网络模型更适合车辆系统搭载,为车辆运动控制提供了实时准确的路面信息。     

用于不平路面车辆动力学仿真的轮胎模型综述

介绍了轮胎在不平路面的动力学特性。在回顾不平路面轮胎动力学模型发展的基础上，以近期的研究工作为重点，对用于不平路面车辆动力学仿真的轮胎模型进行了较为系统的介绍。概要地阐述了各种轮胎模型的建模理论、方法，并进行了分析和评述。最后，总结了不平路面轮胎力学建模的核心问题及发展方向，对不平路面车辆动力学仿真选择合适的轮胎模型给出了建议。     

Model-Based Road Friction Estimation

The tire/road friction coefficient, μ, has a significant role in vehicle longitudinal and lateral control, and there has been associated efforts to measure or estimate the road surface condition to provide additional information for stability augmentation systems of automobiles. In this paper, a model based road friction estimation algorithm is proposed from easily measured signals such as yaw rate and wheel speed. For the development of the estimator, a low order vehicle model incorporated with simple but effective tire model. Field tests of the estimator using actual vehicle measurements show promising results.     

Robust estimation of maximum tire-road friction coefficient considering road surface irregularity

An accurate estimation of the maximum tire-road friction coefficient may provide higher performance in a vehicle active safety control system. Unfortunately, real-time tire-road friction coefficient estimation is costly and necessitates additional sensors that must be installed and maintained at all times. This paper proposes an advanced longitudinal tire-road friction coefficient estimation method that is capable of considering irregular road surfaces. The proposed algorithm uses a stiffness based estimation method, however, unlike previous studies, improvements were made by suggesting a third order model to solve problems related to nonlinear mu-slip curve. To attain the tire-road friction coefficient, real-time normalized force is obtained from the force estimator as exerted from the tire in the low slip region using the recursive least squares method. The decisive aspect of using the suggested algorithm lies in its low cost and versatility. It can be used under irregular road conditions due to its capability of easily obtaining wheel speed and acceleration values from production cars. The newly improved algorithm has been verified to computer simulations as well as compact size cars on dry asphalt conditions.     

Effective assessment of tyre–road friction coefficient using a hybrid estimator

Vehicle stability and active safety control depend heavily on tyre forces available on each wheel of a vehicle. Since tyre forces are strongly affected by the tyre–road friction coefficient, it is crucial to optimise the use of the adhesion limits of the tyres. This study presents a hybrid method to identify the road friction limitation; it contributes significantly to active vehicle safety. A hybrid estimator is developed based on the three degrees-of-freedom vehicle model, which considers longitudinal, lateral and yaw motions. The proposed hybrid estimator includes two sub-estimators: one is the vehicle state information estimator using the unscented Kalman filter and another is the integrated road friction estimator. By connecting two sub-estimators simultaneously, the proposed algorithm can effectively estimate the road friction coefficient. The performance of the proposed estimation algorithm is validated in CarSim/Matlab co-simulation environment under three different road conditions (high-μ, low-μ and mixed-μ). Simulation results show that the proposed estimator can assess vehicle states and road friction coefficient with good accuracy.     

Sensitivity analysis of side-slip angle observer based on a tire model

We investigated the sensitivity of an observer based on a tire model using simulation in linear and nonlinear regions. In the linear region, we investigated the influence of vehicle speed by doing the same simulation at three speed levels. In the nonlinear region, the simulation condition was set such that the vehicle became unstable. In the linear region, steering input and cornering stiffness have a relatively large effect on the estimation error because these quantities determine tire side force. In the nonlinear region, the road surface’s friction coefficient becomes a crucial factor. In both the regions, the observer is sensitive to yaw rate and longitudinal speed.     

Sensitivity analysis of side-slip angle observer based on a tire model

We investigated the sensitivity of an observer based on a tire model using simulation in linear and nonlinear regions. In the linear region, we investigated the influence of vehicle speed by doing the same simulation at three speed levels. In the nonlinear region, the simulation condition was set such that the vehicle became unstable. In the linear region, steering input and cornering stiffness have a relatively large effect on the estimation error because these quantities determine tire side force. In the nonlinear region, the road surface's friction coefficient becomes a crucial factor. In both the regions, the observer is sensitive to yaw rate and longitudinal speed.     

Estimation of maximum road friction coefficient based on Lyapunov method

With the real time and accurate information of motor torque and rotation speed of the four-in-wheel-motordrive electric vehicles, a slip based algorithm for estimating maximum road friction coefficient is designed using Lyapunov stability theory. Modified Burckhardt tire model is used to describe longitudinal slip property of the tire. By introducing a new state variable, a nonlinear estimator is proposed to estimate the longitudinal tire force and the maximum road friction coefficient simultaneously. With the appropriate selection of estimation gain, the convergence of the estimation error of the tire longitudinal force and maximum road friction coefficient is proved through Lyapunov stability analysis. In addition, the error is exponentially stable near the origin. Finally the method is validated with Carsim-Simulink co-simulation and real vehicle tests under multi working conditions in acceleration situation which demonstrate high computational efficiency and accuracy of this method.     

Sideslip angle estimation based on input–output linearisation with tire–road friction adaptation

An adaptive sideslip angle observer considering tire–road friction adaptation is proposed in this paper. The single-track vehicle model with nonlinear tire characteristics is adopted. The tire parameters can be easily obtained through road test data without using special test rigs. Afterwards, this model is reconstructed and a high-gain observer (HGO) based on input–output linearisation is derived. The observer stability is analysed. Experimental results have confirmed that the HGO has a better computational efficiency with the same accuracy when compared with the extended Kalman filter and the Luenberger observer. Finally, a road friction adaptive algorithm based on vehicle lateral dynamics is proposed and validated through driving simulator data. As long as the tires work in the nonlinear region, the maximal friction coefficient could be estimated. This algorithm has excellent portability and is also suitable for other observers.     

Adaptive Yaw Rate Feedback 4WS with Tire/Road Friction Coefficient Estimator

This paper proposes an adaptive yaw rate feedback control system for a four-wheel-steering (4WS) vehicle which involves a tire/road friction coefficient estimator. The adaptive 4WS system has been developed so that the vehicle possesses desirable lateral characteristics even on slippery roads and in critical driving situations. The friction coefficient is estimated on real time from the yaw rate response of the controlled vehicle with the least squares. The control system adopts a two degree of freedom structure which consists of a feedforward compensator and a feedback control subsystem. The feedforward compensator is determined with the estimated friction coefficient to minimize the steady-state and transient vehicle slip angle in spite of changes in tire/road conditions. The feedback subsystem adopts the Internal Model Control (IMC) structure in order to compensate for nonlinearities and to realize robustness against modelling and estimation errors.     

Analytical dynamic tire model

Modeling of tire friction is one of the central problems for vehicle control systems design. LuGre-type dynamic tire model has been proposed and well discussed in previous studies, because it offers a compact form of dynamic model that is convenient in advanced control studies. It has been successfully used in tire slip control design and vehicle state estimation problems. In this article, a concept of time-constrained Stribeck effect is introduced to interpret the mechanism of the LuGre friction model in predicting tire friction characteristics. A modified two-dimensional (2D) dynamic LuGre friction model is introduced to make it compatible with the governing theorem in the steady state. An analytical 2D modified LuGre-type dynamic tire model is developed, in which some fundamental limitations of classical LuGre models are eliminated. The main modifications involve a change in the structure of the 2D LuGre friction model, introduction of load-dependent parameters in 1D and 2D tire models, and a changed structure in the distributed parameter model. The proposed model is compared, in the steady state, to both the Magic Formula and the classical LuGre model. It improves model accuracy in the steady state and gives a physically reasonable distribution of the bristle deflection. A first-order lumped parameter (LP) nonlinear model, which has simpler structure than the distributed parameter model and the classical LP LuGre model, is then derived. Numerical simulations show that the proposed LP model has a good estimation for tire transient dynamics. Thus, the proposed model retains the merits of LuGre-type models and improves the agreement with observation and experimental data on friction force distribution along the patch and on the steady-state friction prediction.