Different Local Model System. With Wiener State Fusion And Its Application

Multisensor information fusion is also called multisensor data fusion, it means to process the data from many sensors data with various and multistage in order to produce newly significant information, and the new information are not obtained from any single sensor. It avoids the single sensor limitation, and can obtain more information to get more accurate and reliable conclusion. Information fusion technology involves the overlapping and concrete utilization of many subjects, such as mathematics, military science, computer science, automatic control theory, artificial intelligence, communication, management science, and so on. It uses the complementation of pleophyletic data and highly speed operation performance of the computer fully, and improves the information processing quality effectively.For linear discrete time-invariant stochastic systems with different local models and multi-sensor, based on the autoregressive moving average (ARMA) innovation model and white noise estimation theory, and combine the modem time series analysis method and classical Kalman filtering method, two kinds of local Wiener state estimators are presented, respectively. For every local state estimator, according to three optimal fusion rules weighted by matrices, diagonal matrices and scalars, the three optimal weighted fusion Wiener common state estimators are presented. Their accuracy is higher than that of each local estimator, and they can handle the fused filtering, prediction, and smoothing problems in a unified framework. In order to compute the optimal weights, the formulas of computing the cross-covariances among local estimation errors are proposed. They can be applied to solve the state fusion filtering problems for the system with color observation noises. By the augmented state approach, the signal to be estimated can be viewed as a common state of the subsystems, so that the information fusion Wiener estimators and Wiener deconvolution estimators are presented for the multisensor ARMA signals with white and color observation noises. Many Monte-Carlo simulation examples in tracking systems and the numerical Monte-Carlo simulation examples show their effectiveness.