基于遗传算法和拟牛顿法的车辆动力学平衡点混合求解方法  

Vehicle Dynamics Equilibriums Solution Search Based on Hybridization of Genetic Algorithm and Quasi-Newton Method

作　　者：王宪彬[1,2] 施树明[1] 刘丽 金立生[1] 

机构地区：[1]吉林大学交通学院,长春130025 [2]东北林业大学交通学院,哈尔滨150040 [3]中国第一汽车股份有限公司技术中心,长春130011

出　　处：《机械工程学报》2014年第4期120-127,共8页Journal of Mechanical Engineering

基　　金：国家自然科学基金资助项目(50775094)

摘　　要：车辆动力学平衡点是汽车非线性动力学分析中的关键参数。但车辆系统的非线性特征使传统的解析方法无法求解平衡点,而基于梯度的数值方法和遗传算法在求解精度和求解效率都存在局限性。为准确有效地求解车辆动力学平衡点,提出一种基于实数编码遗传算法和拟牛顿法的混合求解方法。利用达芬方程验证所提出的方法的有效性。给出基于轮胎魔术公式的3自由度车辆动力学系统模型,利用提出的混合求解方法得到不同转角输入下车辆平衡点,并与单独的遗传算法求解的结果进行对比。研究结果表明,该混合求解方法得到的结果更加准确,该方法能够快速、有效地求解3自由度车辆动力学平衡点。Vehicle dynamics equilibrium is the key parameter in nonlinear dynamics analysis of vehicle stability. Due to nonlinear features of vehicle dynamics, the traditional analytic method, gradient numerical method and genetic algorithm（GA） can hardly work effectively. A hybrid method is proposed for vehicle dynamics equilibrium search, which combines real-coded genetic algorithm with Quasi-Newton（QN） gradient method. The hybrid method is verified by equilibriums solution search of Duffing equation. Based on the magic tyre formula, a three degrees of freedom（3DOF） nonlinear mechanical vehicle model is founded. Numerical experiments with variable front steering angles are carried out to compare proposed GA＋QN with conventional GA on 3DOF vehicle dynamics equilibrium solutions. Numerical results show that the hybrid method can find optimal solutions with higher accuracy and efficiency in solution search for 3DOF vehicle dynamics equilibriums.

关 键 词：车辆动力学平衡点 遗传算法 拟牛顿法 混合求解 

分 类 号：U46[机械工程—车辆工程]