Optimized fast desensitized Kalman filter

The desensitized Kalman filter (DKF) provides an innovative solution for system with uncertain parameters. Acquiring a suitable sensitivity-weighting matrix, which serves as a crucial parameter for balancing the covariance and sensitivity of state estimation error in the DKF, continues to be an unresolved issue. To address the limitation, this paper proposes an optimized fast desensitized Kalman filter (OFDKF) for linear systems, and an optimized fast desensitized extended Kalman filter (OFDEKF) for nonlinear systems based on the PSO. To adjust the sensitivity-weighting matrix, a scaling factor that is designed. A fitness function including the scaling factor is designed for the PSO. An optimal value is obtained by minimizing the fitness function through particle swarm cooperative behavior. Three cases with uncertain parameters are used to verify the proposed OFDKF/OFDEKF have higher estimation accuracy and better robustness for both linear and nonlinear systems.