Robust Parameter Identification Subject to Deterministic Disturbances

遭受定型干擾下之強韌參數識別

10.29979/JCSME.201110.0007

李俊賢(Jyun-Sian Li)；陳明新(Min-Shin Chen)

robust identification ； parameter estimation ； system identification ； Fourier series

中國機械工程學刊

32卷5期（2011 / 10 / 01）

429 - 438

英文

本論文提出一強韌參數識別演算法，使量測輸出在遭受定型干擾下仍能估測出在線性迴歸模型下的未知參數。於之前相關研究中，通常在面臨未知量測干擾時，無法確保能準確估測出系統參數。本文提出以傅立葉餘弦級數來表示干擾訊號，以此作基礎所得到的演算法可獲得準確的參數估測，而無須使用及獲得特殊的擾動模型及資訊。而使用傅立葉餘弦級數法也避免了在使用其他基底函數法所遇到的缺點。最後針對回歸向量的持續刺激性提出其於頻域所需滿足之必要條件。此必要條件將可用來當作設計識別程序時的參考原則。

This paper presents a robust identification algorithm to estimate unknown parameters in a linear regression form that is contaminated by a deterministic disturbance signal. Previous works on robust identification can not guarantee correct parameter estimation when facing unknown disturbances. This paper proposes using the Fourier cosine series to represent the disturbance signal, and the resultant identification algorithm ensures correct parameter estimation without using any disturbance model information. The proposed Fourier cosine series approach is able to avoid many shortcomings of previous basis function approach. Finally, a necessary condition in the frequency domain is proposed for persistent excitation of the expanded regressor. This necessary condition can be used as a guideline for the design of identification process.