Separate-bias estimation scheme with diversely behaved biases

A new separate-bias estimation scheme for systems with diversely behaved biases is proposed. The proposed scheme deals with zero-bias, constant-bias, and random-bias simultaneously to cover diverse behaviors of unknown biases. To efficiently describe the system with these biases, only two system models are developed. Due to fast estimation performance and good inherent properties, two receding horizon Kalman FIR (RHKF) filters are used for corresponding system models to estimate states and biases. It is shown that the constant-bias does not degrade the performance of the state estimation. It is also shown that the bias estimate is obtained independently from states. A detection rule is developed in order to indicate the presence of biases and select the valid estimate from two filters. Via a numerical simulation on the spacecraft attitude tracking system, the performance of the proposed scheme is evaluated by the comparison with the existing Kalman filtering approach.