诱发脑电的ARX参数模型辩识

On ARX Modeling of Evoked Brain Potential

本文分析了诱发脑电的ARX模型(带外输入的自回归模型),指出其参数的最小二乘估计(LS)的有偏性。提出分步辩识方法,充分利用刺激前脑电信号,可得到参数的无偏、一致有效估计。用计算机仿真检验了理论分析。然而,在ARX模型下,模型参数的无偏估计并不保证诱发脑电波形估计的无偏性,我们推荐一个新的模型,可以得到诱发脑电波形的无偏估计。

This paper analyses the bias of Least Squares estimation of ARX model of Evoked Brain Potential (EP). Then, a two—step identification is proposed, making full use of pre—stimuli EEG signal, to get the unbiased, consistent, efficient estimation of model parameters. Computer simulation is Presented. However, in the ARX model of EEG—EP, unbiased model parameters estimation does not guarantee the unbiased EP wave form estimation. A new EEG—EP parametric model is recommended which can attain the unbiased EP waveform estimation.