【作者】 张振宇；

【导师】 何智成； 高晖；

【作者基本信息】 湖南大学 ， 车辆工程（专业学位）， 2021， 硕士

【摘要】 随着汽车技术的不断发展,车辆稳定性越来越受到工程师的关注。对于车辆稳定性的研究,需要借助传感器获取汽车的实时状态信息,如质心侧偏角,但由于汽车开发成本的限制,目前采取的措施是根据可以获取的车辆状态信息,采用估计方法对不易测量的信息进行估算。常见的汽车状态估计方法有滑模观测器、龙伯格观测器、非线性观测器和卡尔曼滤波器等,其中卡尔曼滤波及其改进型算法因为估算稳定精度高的特点成为科研工作者的研究热点。针对车辆动力学状态参数估计问题,本文进行估计算法方面的研究。首先,本文搭建了7自由度整车模型和3自由度单轨模型,构建了整车动力学模型的基础。在整车模型的基础上,本文又分别搭建了魔术轮胎(Magic Tire)和Dugoff轮胎模型。为验证所搭建模型的准确性,本文利用了整车动力学仿真软件Car Sim进行仿真,结果表明,本文所采用模型可以有效估算轮胎力。其次,针对质心侧偏角估算问题,分别建立了基于状态观测器、滑模观测器和扩展卡尔曼滤波(EKF)的估算模型,仿真结果验证了EKF的优越性。针对纵向车速问题,建立了基于递推最小二乘法(RLS)的估算模型,仿真表明RLS法表现比最大轮速法更好。针对路面附着系数估算问题,分别建立了基于附着余量法和EKF的估算模型,仿真结果表明EKF法波动小,更稳定。接着,考虑到EKF在计算非线性系统问题时误差较大的问题,本文采用了容积卡尔曼滤波(CKF)的方法来估算质心侧偏角,并且针对CKF的误差协方差矩阵可能会非正定致使滤波中断的问题,将QR分解用于误差协方差矩阵的计算,解决了矩阵非正定的问题。然后本文建立了基于SRCKF的质心侧偏角估算模型,仿真结果表明SRCKF算法精度高于CKF。最后,利用车辆进行实际路面试验,采集移线工况、绕圆工况和蛇形工况的试验数据,对EKF、CKF和SRCKF离线仿真。试验及仿真结果验证所采用的SRCKF算法的有效性、可行性和准确性。更多还原

【Abstract】 With the continuous development of automotive technology,the stability of vehicles has attracted more and more attention from engineers.For the research of vehicle stability,it is necessary to use sensors to obtain real-time state information of the vehicle,such as the side slip angle of the center of mass.However,due to the limitation of the cost of automobile development,the current measures are to use estimation methods to estimate the information that is not easy to measure based on the available vehicle status information.Common vehicle state estimations include Sliding Mode Observer,Luenberger Observer,Nonlinear Observer,Kalman Filter,etc.Among them,Kalman filter and its improved algorithm have become the research focus of scientific researchers because of the high accuracy of estimation stability.Aiming at the problem of vehicle dynamics state parameter estimation,this paper conducts research on estimation algorithms.First,the 7-degree of freedom vehicle model and the 3-degree of freedom vehicle model are built,which are the basis of the vehicle dynamics model.On the basis of the vehicle model,Magic tire and Dugoff tire models are built respectively.In order to verify the accuracy of the built model,this article uses the vehicle dynamics simulation software Car Sim for simulation.The results show that the model used in this article can effectively estimate the tire force.Secondly,aiming at the problem of estimating the body sideslip angle,the estimation models based on the state observer,SMO and EKF are established respectively,and the simulation results verified the superiority of EKF.Aiming at the longitudinal vehicle speed problem,an estimation model based on the recursive least square method(RLS)is established.The simulation shows that the RLS method performs better than the maximum wheel speed method.Aiming at the problem of road adhesion coefficient estimation,the estimation models based on adhesion margin method and EKF are established respectively.The simulation results show that the EKF method has small fluctuations and is more stable.Next,considering the large error of EKF when calculating nonlinear system problems,this paper uses the square root cubature Kalman filter(SRCKF)method to estimate the body sideslip angle.Aiming at the problem that the error covariance matrix of CKF may be non-positive definite and leading to the interruption of filtering,QR decomposition is used to calculate the error covariance matrix to solve the problem of non-positive definite matrix.Then this article established the estimation model of side slip angle based on SRCKF centroid.Finally,this article uses the vehicle to carry out the actual road test,collects the test data of the shifting working condition,the circle working condition and the serpentine working condition,and simulates the EKF,CKF and SRCKF offline.Test and simulation results verify the validity,feasibility and accuracy of the SRCKF.更多还原