多通道信号自校正分布式融合Kalman滤波器

Kalman filter of self-tuning distributed fusion for multi-channel signal

对带未知参数的多传感器多通道自回归滑动平均(ARMA)信号,采用多维递推辅助变量(MRIV)方法得到自回归模型参数估值,通过Gevers-Wouters算法辨识滑动平均模型参数估值,再用相关方法得到噪声方差的估值。把所有的估值都代入到最优分布式融合信息滤波器中得到自校正分布式融合Kalman信息滤波器。该滤波器具有渐近全局最优性,一个多通道信号仿真例子验证了其有效性。

For multisensor and multi-channel autoregressive moving average（ARMA）signal with unknown parameters,its autoregressive model parameter estimators are obtained by multi-dimensional recursive instrumental variable algorithm. The moving average model parameter estimators are identified through Gevers-Wouters algorithm,and the noise variances estimators are ob-tained with the correlation method. All of the estimators are substitute into optimal distributed fusion information filter to obtain self-tuning distributed fusion Kalman information filter,which has asymptotic global optimality. The effectiveness is verified through a multi-channel signal simulation example.