A Factor Graph Approach For Satellite Attitude Determination By Using Star Sensor

Star sensor has gradually become a necessary sensor for high-performance microsatellite attitude determination system because of its high precision and light weight.However,the star sensor faces shortcomings such as low measurement rate and sensitivity to the spatial environmental impact.Therefore,it is an urgently need for current satellite control research area to explore the multi-sensor attitude determination based on star sensor.The factor graph can visually characterize the relationship between the measurement information and errors of different sensors,and the relationship and influence degree between the past states and current states,and can process the relationship in increment form.The factor graph provides an effective technical approach for deep integration of various sensors.In the framework of the factor graph,this thesis studies the integrated attitude determination by using the star sensor and the gyroscope,and the attitude determination by using star sensor alone.The following research results have been obtained.In order to compensate for the low output frequency of the star sensor,a star sensor/gyroscope integrated attitude determination method based on factor graph is proposed.Firstly,error quaternion based state equation,the measurement equation and the prediction of attitude are used to establish the factor node model,and then sensor/gyroscope integrated attitude determination factor graph model is established.The Gauss-Newton method is used to optimize the factor graph model for achieving satellite attitude calculation.Since the factor graph takes into account the various influencing factors between the sensors,the effective fusion of the star sensor and the gyroscope can be effectively realized.Mathematical simulation results illustrate the effectiveness of the proposed algorithm.Aiming at the problem of large computational complexity by using all measurement information in the optimization of factor graph model,a star sensor/gyroscope integrated attitude determination factor graph method based on incremental smoothing is proposed.Firstly,the Gram-Schmit orthogonalization is used to perform QR decomposition on the Jacobian matrix,and then the factor variance is used to weight the objective function.The incremental characteristics of the factor graph are fully utilized,and only a small number of matrices splicing and replacement are required for optimization of factor graph,which significantly reduces the computation complexity.Aiming at the problem of single-star sensor attitude determination in the absence of gyroscope,a single-star sensor attitude determination method based on factor graph is proposed.In the Modified Rodrigue Parameter framework,the satellite attitude,angular velocity and spatial disturbance torque are used as the state nodes,and the factor node model is built by using satellite dynamics,star sensor measurement equation and noisedriven spatial disturbance torque.In this way,the single-star sensor attitude determination factor graph is established.Then,the objective function is weighted by the factor variance and retaining parts of the past states in the optimization.Finally,the factor graph is optimized by using the incremental smoothing method.The mathematical simulation results demonstrate that the proposed method can effectively realize the attitude determination by using the single-star alone.The high accuracy of attitude determination can be achieved and the computation complexity is significantly reduced.