本文提出了一种加性有色高斯噪声中因果非最小相位ARMA模型的自适应辨识算法。模型输入假定为非高斯独立同分布随机过程。算法只利用了观测信号的高阶统计量。在每次迭代中,先估计AR参数,再估计MA参数,但不用计算残差序列。在参数递推中采用了随机梯度法。仿真实验证实了本文算法的有效性。 In this paper, we propose an adaptive identification algorithm for causal non-minimum phase ARMA model in additive colored Gaussian noise. The model inputs are assumed to be non-Gaussian independent and identically distributed random processes. The algorithm only uses the high-order statistics of the observed signal. In each iteration, we first estimate the AR parameters and then estimate the MA parameters, but we do not need to calculate the residual sequence. Stochastic gradient method was adopted in parameter recursion. Simulation results show the effectiveness of the proposed algorithm.