Information fusion in Kalman filter weighted by diagonal matrices
按对角阵加权信息融合Kalman滤波器

In order to overcome the drawbacks that the information fusion in non-steady-state Kalman filter weighted by matrices requires a large on-line computational burden,and that the accuracy of the fused Kalman filter weighted by scalars is low,a multisensor information fusion in steady-state Kalman filter weighted by diagonal matrices is presented by the modern time series analysis method.It is equivalent to the information fusion in Kalman filters weighted by scalars for the state components,so that the decoupled information fusion in Kalman filters is achieved.Its accuracy and computational burden are between those weighted by matrices and weighted by scalars.It is suitable for real time applications.In order to compute the optimal weights,the Lyapunov equations for computing the filtering error variance and covariance matrices are also presented.A simulation example for an radar tracking system with three-sensor shows its effectiveness.