摘要
在简单讨论逆模型辨识原理的基础上,利用支持向量机(SVM)对函数逼近的能力,提出了基于支持向量机的直接逆模型辨识方法.分别采用二次核函数以及高斯RBF核函数,利用训练数据对线性和非线性系统进行黑箱辨识.仿真结果表明,基于支持向量机的直接逆模型辨识方法在处理线性和非线性对象时,辨识性能都优于传统的BP神经网络,不仅辨识精度高,辨识速度快,而且泛化能力较强.
After a simple discussion of the principle of the inverse_model identification,a support vector machines(SVM) based direct inverse_model identification method is developed by using SVM's excellent ability of function approximation.According to the train data,linear and nonlinear systems' black_box identification is performed by using SVM with quadric polynomial and Gaussian RBF kernel respectively.Simulation results show that the performance of SVM based direct inverse_model is better than that of BP neural network in that it has better identification precision,quicker identification speed and stronger generalization ability.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2005年第2期307-310,共4页
Control Theory & Applications
基金
国家973项目资助(2002CB312200).
