摘要
本文研究了带未知白色观测噪声的ＡＲ模型多数的无偏估计问题，提出了一种实现ＡＲ模型参数无偏估计的偏差补偿最小二乘法，这种方法通过对观测数据预滤波，将一个已知零点嵌入被辨识系统，并利用该零点提供的信息，从普通最小二乘估计中提取出噪声引起的偏差并予以消除，从而得到无偏估计。文中给出的数值仿真例子说明了所提算法的有效性。
The problem of consistently estunating parameters of AR model in the presence of measur-ins noise is studied. A new bias-eliminating least-squares method is proposed. A known prefilter is in-serted into the system to be identified so that the system has a known zero, which can be ued to ex-tract the noise-induced estimation bias. With the bias eliminated the consistent paranieter estimates areobtained. The simulation results verifies the theoretical analysises.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
1995年第3期67-71,共5页
Journal of Southeast University：Natural Science Edition
