利用SVR的非线性动态系统Wiener模型补偿  

Nonlinear Dynamic Compensation for Wiener Model Using Support Vector Regression

作　　者：吴德会[1,2] 

机构地区：[1]九江学院电子工程系,江西九江332005 [2]合肥工业大学仪器仪表学院,安徽合肥230009

出　　处：《小型微型计算机系统》2008年第3期552-556,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(70272032)资助;江西省教育厅科技项目(2007-328)资助

摘　　要：提出一种新的利用支持向量回归机(SVR)的非线性动态系统维纳(Wiener)模型补偿方法.首先,将非线性动态系统用Wiener模型描述成线性动态子环节和非线性静态增益;再设计结构上与之对应的Wiener补偿器,并进一步将其变换为可用SVR辨识的线性中间模型;最后,通过关系矩阵将中间模型的估计值转换为Wiener补偿器的实际参数.用实际压力响应系统的动态标定实验数据进行测试,结果表明,与最小二乘方法比较,所提方法建立的Wiener补偿器具有更强的抗干扰能力.因此,该研究为非线性动态系统补偿又提供了一种可选方法.A novel nonlinear dynamic compensation method based on Wiener model using support vector regression （SVR） was presented in this paper. Firstly, the nonlinear dynamic characteristic of system was expressed by a linear dynamic subunit followed by a nonlinear static subunit-Wiener model. Then, the nonlinear transfer function of Wiener compensator was converted to linear intermediate model which could be identified by SVR. Finally, the transform matrix was derived, through which coefficients of Wiener compensator were gotten from intermediate model. Practical dynamic calibrating experimental data of pressure-transducer system are used to test, and the results show, compared with least squares （LS） identification methods, the proposed Wiener compensator was more robust in noise resistance. So it provides a better way for dynamic compensation of nonlinear system.

关 键 词：非线性动态系统 支持向量回归机 维纳模型 补偿 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]