基于非线性规划的指数自回归模型的辨识

On Identification of Exponential Autoregressive Model with Nonlinear programming

讨论了指数自回归模型的辨识问题，证明了该模型最小二乘估计的目标函数的非凸性，并给出了使该函数为凸的条件；最后给出了辨识该模型的算法及该算法的收效性，并以数值例子加以说明。

The identification of exponential autoregressive model (EAR modal)is discussed in this paper. It is proved that residual squares sum is not convex and a condition is given under which the residual squares sum is convex. A method of identification on EAR model is given and illustrated with a numerical example.