基于神经网络的多维自回归模型参数估计

Parameter Estimation for Multi-Dimension Autoregressive Models via Neural Network

本文提出一种基于神经网络的多维自回归模型（ＡＲ，ＮＬＡＲ）参数估计方法。该方法采用预畸变技术，使神经网络能适合于多维自回归模型参数估计。其优点是直接利用时域信息、网络结构简单、估计准确。最后给出了一些模拟结果。

A method of parameter estimation for multi-dimension autoregressive models(AR,NLAR)via neural network is given in this paper.The neural network can suit parameters estima-tion for multi-dimension autoregressive models by using pre-distorted technique.The advantagesof the method are time-domain information used directly,simple network constructure and accu-rate estimation. Finally,some simulated results are provided in this paper.