一种收敛于全局无偏最优解的非线性自回归（NAR）预测

本文研究了基于Ｖｏｌｔｅｒｒａ级数展开的非线性自回归预测模型的参数估计问题，从线性自适应ＩＩＲ波滤输出误差算法的观点，研究ＮＡＲ模型的参数估计问题，利用平均收敛条件，提出一可收敛于全局无偏最优解的新算法，解决了一般的预测误差算法的有偏解问题。

A Nonlinear Autoregressor Predictor with the Global & Unbiased Solution

The object of this paper is to investigate the parameters estimation of nonlinear autoregnssor(NAR) predictor. From the view point of adaptive infinite impulse response (IIR) filtering output error algoritlun,the NAR parameters estimation is investigated. With the help of the time averaging convergence condition, a new parameters estimate algorithm is Presented, which converges to a globally unbiased solution even ifthe additive noise exists.