【摘要】 针对于子空间辨识算法辨识闭环系统时,由于输入信号与不可测噪声是相关的,往往会得到有偏估计的问题。提出一种采用自回归滑动平均模型(ARMAX)的闭环子空间辨识方法,通过扩展最小二乘方法(ELS)估计ARMAX模型中的马尔科夫(Markov)参数,使用预测的子空间辨识方法(PBSID)获取系统参数矩阵,避免了采用高阶自回归模型(ARX)所导致的过大的估计方差等问题。算法实例验证结果表明,改进方法能够获得较好的闭环系统一致性估计,辨识精度较高,有非常良好的应用前景。更多还原

【Abstract】 Most of the existing subspace identification methods get biased estimation for closed-loop conditions,because of the input signal and the unmeasured noise are related. This paper presents a closed-loop subspace identification method based on Auto Regressive Moving Average model; estimates Markov parameters in the ARMAX model by the extended least square method; recovers system parameter matrix by using the predictor-based subspace identification method. It avoids the too large estimation variance caused by high order Auto Regressive model. A closed-loop simulation example is given to demonstrate that the identification algorithm can obtain consistent estimates and is effective.更多还原