多传感器多变量AR模型信息融合辨识方法

对于带白色公共干扰噪声、白色观测噪声和传感器偏差的多传感器多变量自回归（AR）模型，当AR模型参数、传感器偏差和噪声方差未知时，提出了一种信息融合多段辨识方法，其中用多重递推增广最小二乘法（MRELS）得到AR模型参数和传感器偏差的局部和融合估值器，再用相关方法得到局部和融合噪声方差估值器。这些估值器具有一致性。一个仿真例子验证了其有效性。

Information fusion Identification Method for Multisensor Multivariable AR Models

For the multisensor multivariable autoregressive(AR) model with white common disturbance noise, and white measure noise, and sensor bias, when the AR parameters, sensor bias, and noise variances are unknown, an information fusion multi-stage identification method has been presented, where the local and fused estimators of the AR parameters and sensor bias are obtained by the multiple recursive extended least squares (MRELS)algorithm, and the local and fused estimators of the noise variances are obtained by the correlation method. These fused estimators have consistency. A simulation example shows its effectiveness.