Fast information fusion Kalman filter
Yuan Gao · Kongzhi yu juece · 2005
By the modern time series analysis method, under the linear minimum variance fusion criterion weighted by scalars, a multisensor fast information fusion steady-state Kalman filter is presented, where the gain is computed via the autoregressive moving average(ARMA) innovation model. And Lyapunov equations are presented for computing the filtering error variance and covariance matrices among sensors, which can be solved by iteration. The exponential convergence of the iterative solution is proved. Compared with the Riccati equation-based information fusion Kalman filter weighted by matrices, it can obviously reduce the computational burden, and is suitable for real time applications, It can be applied to design information fusion self-tuning Kalman filter for systems with unknown noise statistics. A simulation example for a target tracking system shows its effectiveness.