A self-calibration filter based on semi-parameter modeling and DUKF

Compensations or eliminations of the systematical error are indispensable for designing an accurate real-time instrumental system. In virtue of an elaborate object function, a new recursive model for the state and the systematical error is developed based on semi-parameter modeling. Considering of the separable character of the new model, an improved dual unscented filter (DUKF) is developed, which can estimate the state and the systematical error simultaneously. The new algorithm has excellent self-calibration ability for the systematical error as well as a marked accuracy of the state estimation, which are validated by simulations.