首页> 外文学位 >Application of higher-order spectral analysis to nonlinear system identification.
【24h】

Application of higher-order spectral analysis to nonlinear system identification.

机译:高阶谱分析在非线性系统辨识中的应用。

获取原文
获取原文并翻译 | 示例

摘要

In this study, recent advances in higher-order spectral analysis techniques and Volterra system theory have been utilized to yield a practical tool for the identification of weakly nonlinear systems up to third order. The emphasis is on the frequency-domain analysis of the system input-output data because of the insight it gives us into a very powerful technique, i.e., nonlinear transfer function (NTF) approach. The NTF approach is general and practical in that (i) it can handle various types of nonlinear systems (i.e., systems which can be described by a Volterra series up to third-order), and (ii) a wide variety of input excitations (including Gaussian and nonGaussian) can be utilized for the system identification.;A matrix formulation for the system I/O representation is presented. Next, a description of the identification techniques and their digital implementations, which employ the well-known linear solution techniques, e.g., the Cholesky method and a RLS adaptive algorithm, are given in detail. Furthermore, experimental knowledge of the system transfer functions is utilized to (i) solve the spectral decomposition problem, (ii) compute the generalized system coherency functions, which provide a goodness-of-fit measure for the validity of the Volterra model, (iii) estimate the system parameters that appear in many nonlinear system equations, and (iv) reduce unwanted nonlinear effects or linearize a given Volterra system. Also, an efficient and fast algorithm for the classification of the third-order intermodulation and harmonic products is presented, which utilizes the analysis results derived for the NTF approach.
机译:在这项研究中,高阶频谱分析技术和Volterra系统理论的最新进展已被利用来提供一种实用的工具,用于识别直至三阶的弱非线性系统。重点是对系统输入输出数据的频域分析,因为它使我们了解了一种非常强大的技术,即非线性传递函数(NTF)方法。 NTF方法是通用且实用的,因为(i)它可以处理各种类型的非线性系统(即,可以用Volterra级数描述的系统,直到三阶),以及(ii)各种各样的输入激励(包括高斯和非高斯)可用于系统识别。;提出了用于系统I / O表示的矩阵公式。接下来,详细描述识别技术及其数字实现,该技术采用众所周知的线性解
决技术,例如,Cholesky方法和RLS自适应算法。此外,利用系统传递函数的实验知识来(i)解决频谱分解问题,(ii)计算广义系统相干函数,这为Volterra模型的有效性提供了拟合优度度量,(iii )估计出现在许多非线性系统方程式中的系统参数,并且(iv)减少不必要的非线性影响或线性化给定的Volterra系统。此外,提出了一种高效,快速的三阶互调和谐波产物分类算法,该算法利用了NTF方法得出的分析结果。

著录项

  • 作者

    Nam, Sang-Won.;

  • 作者单位

    The University of Texas at Austin.;

  • 授予单位 The University of Texas at Austin.;
  • 学科 Electrical engineering.;Mechanical engineering.;Engineering.
  • 学位 Ph.D.
  • 年度 1990
  • 页码 149 p.
  • 总页数 149
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号