Robust Centralized Fusion Kalman Filtering Under Uncertain Noise Variance and Multiple Networked Inducements

The robust centralized fusion (CF) filtering problem for a system with linearly correlated noise and mixed uncertainties of noise variances, multiplicative noises, and multiple networked inducements, including missing measurements, packet dropouts, and two-step random measurement delays (RMD), is addressed in this paper. First, the model transformation is performed for the centralized architecture of the original system by the integrated model transformation method. The transformed system has comparatively fewer dimensions. Then, the CF time-varying recursive robust Kalman filter is presented based on the minimax robust filtering method and the decorrelation method. The extended Lyapunov equation approach is presented to prove that the actual filtering error variance is guaranteed to have minimal upper bounds for all admissible uncertainties. Next, the accuracy relations and convergence in a realization are proved. The CF steady-state robust Kalman filter is also presented. Finally, a simulation example applied to the F404 aircraft engine system is to demonstrate the effectiveness of the proposed results.