融合状态观测及优化方法的纯侧偏轮胎模型辨识  

Tire pure sideslip model recognition using fused state observation and optimization method

作　　者：邱香[1] 吴晓建 周兵[3] QIU Xiang;WU Xiaojian;ZHOU Bing(Collaboration Innovation Center,Jiangxi University of Technology,Nanchang 330098,China;School of Mechanical&Electrical Engineering,Nanchang University,Nanchang 330031,China;State Key Lab of Advanced Design and Manufacture for Vehicle Body,Hunan University,Changsha 410082,China)

机构地区：[1]江西科技学院协同创新中心,南昌330098 [2]南昌大学机电工程学院,南昌330031 [3]湖南大学汽车车身先进制造国家重点实验室,长沙410082

出　　处：《振动与冲击》2020年第13期84-90,共7页Journal of Vibration and Shock

基　　金：国家重点研发计划新能源汽车重点专项项目子课题(2016YFB0100903-2);国家自然科学基金项目(51875184,51765021);江西科技学院开放基金项目(16XTKFYB05)。

摘　　要：针对目前轮胎模型辨识大多基于轮胎力、轮胎侧偏角等数据已知或可测假设的局限,提出一种基于车载传感器和整车操稳性试验的魔术公式轮胎纯侧偏模型辨识方法。该方法融合状态观测与优化思想,首先构建无迹卡尔曼滤波(Unscented Kalman Filter, UKF)状态观测系统对轮胎模型特征参数进行初步估计,而后将参数识别转换为优化问题,由UKF状态观测系统为粒子群优化(Particle Swarm Optimization,PSO)算法提供初值,且将UKF估计结果与魔术公式轮胎纯侧偏模型参数灵敏度分析结果相结合,为PSO算法提供搜索区间,进一步获取更精确辨识结果。Simulink仿真及不同侧向加速度下的Simulink-Carsim联合仿真结果共同表明,车辆侧向加速度达到一定程度使轮胎进入非线性域后,所提出的辨识方法能够获得较准确的辨识效果。Aiming at limits of most current tire model recognition being based on the assumption of tire force,tire sideslip angle etc.being known or measurable,a new method for tire pure sideslip model recognition based on the magic formula(MF)of on-board sensor and overall vehicle handling stability test was proposed.In this method,the state observation and optimization method were fused.Firstly,characteristic parameters of the tire model were preliminarily estimated using the unscented Kalman filter(UKF)state observation system,and then the parametric recognition was converted into an optimization problem.the UKF state observation system provided initial values for the particle swarm optimization(PSO)algorithm.UKF estimation results combined with parametric sensitivity analysis ones gained with MF of tire pure sideslip model to provide a search interval for PSO algorithm,and further obtain a more accurate recognition.The results of Simulink simulation and Simulink-Carsim joint simulation under different lateral accelerations showed that when lateral acceleration of a vehicle reaches a certain level to make tire enter nonlinear domain,the proposed recognition method can get more accurate recognition results.

关 键 词：魔术公式轮胎模型 参数辨识 状态观测 无迹卡尔曼滤波 粒子群优化 

分 类 号：U461.1[机械工程—车辆工程]