Covariance intersection fusion Kalman predictor for two-sensor descriptor system

For the two-sensor linear stochastic descriptor system, the covariance intersection (CI) fusion Kalman predictor is presented, in order to improve the prediction accuracy. The two-sensor descriptor system can be transformed into two reduced-order non-descriptor subsystems by the singular value decomposition(SVD) method, and the local Kalman predictors of the two-sensor descriptor system are obtained applying the classical Kalman filtering. Then, the CI fusion Kalman predictor and its prediction error variance of the two-sensor descriptor system are presented, which can avoid computing the cross-covariances of the local predictors. It is proved that its accuracy is higher than that of each local Kalman predictors, and is lower than that of the fusion predictor weighted by matrices and weighted measurement fusion (WMF) Kalman predictor. A simulation example verifies the effectiveness.