对角双线性模型辨识基础与谱分析  

Identification theory and spectral analysis of diagonal bilinear models

作　　者：王海斌[1] 韦博成[1] 陈浩球[1] 

机构地区：[1]东南大学应用数学系,南京210096

出　　处：《系统工程学报》2002年第2期126-132,192,共8页Journal of Systems Engineering

基　　金：国家自然科学基金资助项目 ( 196 310 40 );江苏省自然科学基金资助项目 ( BK990 0 2 )

摘　　要：由于能对非线性系统很好的近似 ,双线性模型在许多工程领域已得到了广泛的应用 .但其统计基础还很不完善 .对于对角双线性模型 (DBL(p,q,r) m odels) ,利用多元 Wiener- Ito随机积分 ,计算出了输入输出序列的自协方差、互协方差和双互协方差 .在此基础上进行了谱分析 ,得到了该模型的谱密度、互谱密度和双互谱密度 .这些结果可以用作该模型的辨识、预报和控制 .As a kind of efficient approximation to nonlinear system, bilinear models have a widely applications in many engineer fields. However, Statistical theory of bilinear model is rather limited. By means of multivariate Wiener-Ito stochastic representation, diagonal bilinear model (DBL model) had been discussed. Auto-covariance and cross-covariance and bi-cross-covariance of DBL model are evaluated individually in time domain. Spectral density and cross-spectral density and bi-cross-spectral density are also derived in frequency domain. These results can be used for identification prediction and control of DBL models.

关 键 词：对角双线性模型 辨识 谱分析 非线性系统 

分 类 号：TP271[自动化与计算机技术—检测技术与自动化装置]