ARMA模型的参数估计,是一个非线性寻优的过程,本文提出一种新的神经网络算法,用于ARMA模型的参数估计,可以保证计算的收敛性,提高运算速度和估计精度。 ARMA model parameter estimation is a non-linear optimization process. In this paper, a new neural network algorithm is proposed for parameter estimation of ARMA model, which can ensure the convergence of computation and improve the computing speed and accuracy.