Robust Centralized Fusion Steady-State Kalman Predictor with Uncertain Parameters

For multisensor time-invariant systems with uncertain parameter and known noise variances, the centralized fusion robust steady-state Kalman predictor based on the minimax robust estimation principle is presented by a new approach of compensating the parameter uncertainties by fictitious noise. Using the Lyapunov equation, it is proved that the variances of its actual prediction error variances have a conservative upper bound when the uncertainty of parameters is restricted in a sufficiently small region, which is called the robust region of the parameter uncertainties. It is also proved that the robust accuracy of the centralized fuser is higher than that of each local robust Kalman predictor. A simulation example shows how to search the robust region and shows its good performances.