引言在二阶平稳时间序列的混合自回归动平均(ARMA)的表达式中,现在的观测可表示为它自己的过去历史的有限线性组合(自回归部分)加上不能预测的部分,后者是由不相关随机变量的有限线性组合(移动平均部分)构成的。本文讨论只利用输出数据时对已知阶数的混合ARMA 时间序列的自回归(AR)参数的估计问题。它等效于对于一个平稳线性离散时间系统, INTRODUCTION In the ARMA expression of second-order stationary time series, the present observations can be expressed as a finite linear combination (autoregressive) of its own past history plus an unpredictable fraction, the latter Is composed of a finite linear combination (moving average) of uncorrelated random variables. This paper discusses the problem of estimating the autoregressive (AR) parameters of a mixed ARMA time series of known orders using only the output data. It is equivalent to a stationary linear discrete-time system,