摘要
提出一种新的尺度核支持向量回归方法,并应用于非线性系统辨识问题。通过线性规划技术和采用尺度函数作为核函数来实现支持向量回归模型。该支持向量回归的核函数不必满足Mercer条件,为实际应用更广泛地选择尺度核提供更大的灵活性。仿真实验结果表明该尺度核支持向量回归方法可以成为非线性系统辨识的有力工具。
A new scaling kernel support vector regression was proposed for nonlinear system identification problem. Using linear programming technique and scaling kernel function, the support vector regression model was obtained. The kernel function of support vector regression doesn＇t need to meet Mercer condition so as to offer more flexibility for selecting support kernel in practice application. The simulation results show that the scaling kernel support vector regression method can become the powerful tool for the nonlinear system identification.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2006年第9期2429-2432,共4页
Journal of System Simulation
基金
教育部博士点培养基金资助项目(20040613013)
四川省教育厅重点项目(0229957
2005A117)
关键词
非线性系统辨识
小波理论
支持向量回归
尺度核支持向量回归
Nonlinear system identification
Wavelet theory
Support vector regression
Scaling kernel support vector regression
