基于滤波技术的随机连续系统ELS辨识及其数值实现*

ELS Identification and Its Numerical Realization for Stochastic Continuous Systems Based on Filtering Technique

中文关键词  随机连续系统  系统辨识  最小二乘法  常微分方程数值解  Wiener过程

英文关键词  stochastic continuous systems  system identification  least squares method  numerical solu-tion of ordinary differential equations  Wiener process

中文摘要

      本文对低阶扰动模型，讨论滤波器的选取，从而通过对输入输出滤波，使辨识随机连续系统的连续ELS（增广最小二乘）法一致收敛的SPR条件得到满足，基于数值积分和常微分方程数值解，文中还讨论ELS辨识的两种数值实现. 仿真结果显示本文方法的有效性.

英文摘要

      For the low-order disturbance model， how to choose the filter is discussed in this paper， so that the SPR condition of the consistently convergence of the continuous ELS identification for stochastic continu-ous systmes based on filtering of the input and output is satisfied. Then two kinds of numerical realizations of the ELS method are given based on the numerical integral and the numerical solution for ordinary differential equations. Simulation results show the effectiveness of these realization methods.