基于神经网络的时序预测模型研究

将神经网络作为传统的时序线性模型的非线性推广进行了分析，论证了多层前向神经网络与非线性自回归模型及反馈神经网络与非线性自回归移动平均模型的等价意义，提出了一种可作为非线性时序模型的内反馈神经网络．

Research on Time Series Prediction Model Based on Neural Networks

In this paper, neural network is analyzed as the nonlinearextension of traditional linear time series models, and it is demonstrated thatmultilayered feedforward neural network is equivalent to nonlinear autoregressive model,and recurrent neural network is equivalent to nonlinear autoregressive moving averagemodel. At last, a kind of internal recurrent neural network is proposed as it may be thenonlinear time series model.