Information fusion multi-stage identification method for multisensor multi-channel ARMA models

For multisensor multi-channel autoregressive moving average(ARMA) signal with white measurement noises and a common disturbance measurement noise, when the model parameters and the noise variances are all unknown, an information fusion multi-stage identification method is presented. It consists of three stages: In the first stage, the local and fused estimates of the autoregressive(AR) parameters are obtained by the multiple recursive instrumental variable (RIV) algorithm. In the second stage, based on the sampled cross-correlation functions, the local and fused measurement noises variances estimates are obtained. In the third stage, applying the Gevers-Wouters algorithm with a dead band, the local and fused estimates of the moving average(MA) parameters and process noise variances are obtained. The fused estimates are obtained by taking the average of the corresponding local estimates. The consistency of the fused estimates is proved. One simulation example is given to verify the consistency of the estimates.