Research On Attitude Detemination Algorithm Of Fast Maneuver Satellites

The nonlinearity of the satellite model is so high when the satellite is in thecondition of fast maneuver that ordinary attitude determination algorithm will causea respectively huge model error which will make the precision of the filter descenddrastically. For this condition, this thesis introduces several estimation methods anddesigns attitude determination algorithm based on different model assumptions andestimation principles. Furthermore, this thesis designs the fault detection andisolation algorithm for the data of star sensor is easily invalid when the satellite ismoving fast. The contents of this thesis include:Introduce the basic knowledge of satellite attitude determination algorithmbriefly. Establish the satellite’s maneuver model and dynamic model. Establish theobservation equation of attitude sensors which are used in this article.Introduce the Extend Kalman Filter (EKF) algorithm, linearization the systemmodel and establish the algorithm. As the linear model differs too much from thereal model, this paper introduces the Nonlinear Predictive Filter (NPF) theory,assumes there is a model error in the linear system. Base on the Minimize ModelError (MME) principle, this paper revises the model error by using the observationdata to make the linear model approach the real model, therefore gets an accurateestimation of the system state by using the revised model.This thesis introduces the Unscented Kalman Filter (UKF) algorithm as thesystem is highly nonlinear. The UKF does not do any transformation on the systemmodel, therefore it is adaptive to the nonlinear systems. Because of the hugecomputation, this paper discusses the Spherical Simplex Unscented Transformation(SSUT), consequently the number sampling points descends sharply. This algorithmreduces the computation drastically and maintains the original accuracy.Introduces the robust filter algorithm, treat the model error as the system noisewhich has the limited energy. As there are several uncertain factors in the model, thealgorithm cannot get the error variance matrix directly. Thus this article discussesthe matrix inequality and gets the estimation algorithm based on principle of theminimized upper limit of the error variance matrix. Finally this paper establishes theattitude determination algorithm for the system model which includes uncertainfactors.This article introduces the Strong Tracking Filter (STF) algorithm because thesystem state changes quickly when the satellite is moving fast. This algorithm enforces the observation error have the character of Gauss Noise by using theOrthogonal Principle, and fully extract the information of the observation, thus thisalgorithm can realize the tracking on real system state.As the star sensor is easily fault when the satellite is moving fast, this paperdiscusses the fault detection and isolation algorithm for the star sensor. Thisalgorithm can isolate the star sensor when it’s data is invalid, and relieve theisolation when the star sensor recovers.