最小二乘支持向量机在汽车动态系统辨识中的应用  

Application of LS-SVMs in the Automobile Dynamical System Identification

作　　者：郑水波[1] 韩正之[1] 唐厚君[1] 张勇[2] 

机构地区：[1]上海交通大学电子信息与电气工程学院,上海200030 [2]上海交通大学机械与动力工程学院,上海200030

出　　处：《上海交通大学学报》2005年第3期392-395,共4页Journal of Shanghai Jiaotong University

基　　金：上海汽车工业科技发展基金会项目(0224)

摘　　要：汽车转向时动态系统参考模型对于汽车稳定性的控制有重要影响.基于最小二乘支持向量机算法,应用网络搜索和交叉验证的方法选择支持向量机参数,并将其应用于汽车转向时的非线性动态系统辨识,取得了良好的辨识效果,建立的参考模型能够较充分地描述汽车动力学行为.The dynamical desired model has an important impact on the vehicle stability control during cornering. A nonlinear modeling method based on Least Squares Support Vector Machines (LS-SVMs) was proposed. LS-SVMs hyperparameters are optimized with grid-search and cross-validation methods. LS-SVMs is applied to modeling the desired vehicle dynamical system. The research result shows that it has good performance on the identification of nonlinear dynamical system. The desired model can fully represent the automobile dynamics.

关 键 词：最小二乘支持向量机 系统辨识 网格搜索 交叉验证 汽车参考模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]