A Threshold-Based Kalman Filter with Recursive Covariance Estimation

In the existing works, a Kalman filter with recursive covariance estimation (KF-RCE) was proposed by Bo Feng et al. to sequentially estimating process noise covariance matrix. However, there may be a singular matrix existed in the KF-RCE algorithm, which would lead to unreasonable system state estimation in the system initial stage. In this paper, a threshold-based Kalman filter with recursive covariance estimation (TBKF-RCE), which uses a threshold-based switch scheme, is presented to suppress the singularity phenomenon brought by the singularity matrix in KF-RCE algorithm. Simulation results have shown that the singularity phenomenon which happened in the system initial stage has been weakened by using the TBKF-RCE algorithm.