Research On Dynamic Techniques For The Star Sensor With Attitude Correlated Frames Approach

Star tracker,or star sensor,is one kind of the attitude sensors that is widely used in the applications of attitude measurements and calibrations for deep space exploring,satellites,ballistic missiles,aircrafts and ships.However,the attitude accuracy of a star sensor will decrease seriously under highly dynamic conditions.How to improve the dynamic performance of a star sensor has become a research spot in recent years.In this thesis,an attitude-correlated frames approach?ACF?is proposed to improve the attitdue accuracy of the star sensor under highly dynamic conditions.The work of this thesis focused on three aspects,namely the motion-blurred star image processing,the derivation and verification of the ACF approach and its performance analysis,calibration methods of intrinsic and extrinsic parameters of the star sensor.The major research work of the thesis is summarized below.1.Processing methods of the motion-blurred star image,including the removal of complex background noises,segment and extraction of blurred areas and the restoration of the blurred star image.Based on the features of motion-blurred star image and the ad-vantages of the mathematical morphology,the mathematical morphology operations are researched and employed in the removal of the noises including the cluttering background lights,clouds,moonlights,the segment of neighboring stars and the reconstruction of break motion-blurred stars.Simulation and experiments have verified that the mathemat-ical morphology operations are effective in processing the motion-blurred star images.A segment and extraction method of the motion-blurred star image based on template-correlation matching is proposed to obtain the star image areas and reduce the noises as well.Simulation and experimental results show that this method readily identifies and extracts the star spot area and removes the noises outside its region simultaneously.It is effective in extracting faint blurred star images unless they are totally submerged by the noises.A region-confined restoration method,which concentrates on the noise re-moval and signal to noise ratio?SNR?improvement of the motion-blurred star images,is proposed for the star sensor under dynamic conditions.A multi-seed-region growing technique with the non-linear kinematic recursive model for star image motion is given to find the star image regions and to remove the noise.A restoration strategy based on Bayes principle is employed to restore the motion-blurred star image.Simulation results indicate that the region-confined restoration method is effective in removing noise and improving the centroiding accuracy under different noise level and dynamic conditions.The SNR of the star image has improved for 5 to 6 times with the proposed method in the experiment.Identification rate and the average number of identified stars in the experi-ments has improved more than 2 times and about 3 times respectively,which verifying the advantages of the region-confined restoration method.2.The attitude-correlated frames approach and its verification.An attitude-correlated frames?ACF?approach,which concentrates on the features of the attitude transforms of the adjacent star image frames,is proposed to improve upon the existing techniques.The attitude transforms between different star image frames are measured by the strap-down gyro unit precisely.With the ACF method,a much larger star image frame is obtained through the combination of adjacent frames.As a result,the number of effective star im-ages is increased and the random noise is removed,the degradation of attitude accuracy caused by motion-blurring are compensated for.Two simplified ACF approaches,namely the WACF and the RACF,are derived and verified.Comparisons between the ACF and the simplified ones are given,the WACF and RACF rely on the attitude of each single frame,but it they are much more concise.Simulations and experimental results indicate that the ACF approach is effective in removing random noises and improving the attitude determination accuracy of the star sensor under highly dynamic conditions and different noise levels.The improvement of the attitude accuracy is approximately proportional to the square root of the number of correlated star image frames.3.Performance analysis of the ACF approach.Combining the procedure of the ACF approach,the storage and time consumption is analyzed,the results show that it is not a burden for the star tracker to implement the ACF approach.The transform features of the observed vectors are analyzed when the rotation angle of the star sensor is small and big.Gyroscope noises are discussed for the ACF approach,a model is established between the correlation time and the gyroscope noises.The detecting capacity of the ACF approach under different dynamic conditions is researched.Upon the above researches,an optimal correlation time for the ACF approach is?t_c?_max^optimal=min[?/?9?_s?,N_max/f_update].The state equation and observing equation is established for the Kalman Filter between the star sensor and the gyroscope unit.The ACF approach can output the attitude information continuously by increasing the number of frames N and obtain a much better accuracy than the KF method.However,the gyro bias can be estimated by the KF method,which is beneficial for long-term attitude determination.4.Calibration methods of the intrinsic and extrinsic parameters of the star sensor.Upon the research on existing techniques,a calibration method referred to the ACF ap-proach?ACFCM?is proposed.ACFCM method combine with the advantages of the at-titude dependent methods and attitude independent methods.Simulation and experiment results both show that the performance of the ACFCM method is better than the IAICM method,the ACFCM method is reliable and the calibration accuracy of the principle point is 3 times to the IAICM method.A calibration method considering the image star cen-troiding error is proposed.The method is based on the principle of the SPGD algorithm.The optimal calibration parameters are obtained through an iteration procedure in which the disturbance is added to the star centroids.Simulation result indicate that the intrinsic parameters are calibrated out and kept constantly after tens of iterations.The accuracy of principle point of the proposed SPGD based method is 4 times as the IAICM,and is im-proved by 40%comparing to the ACFCM method.As for the focus length,the accuracy is improved by 6 times and 3 times compared to the IAICM and ACFCM respectively.Simultaneously,the calibration operation is simplified.The effect of the mounting an-gle to the ACF approach between the star sensor and the gyro unit is analyzed.Then a calibration method for the mounting angle is proposed based on the equivalent rotation vector.Simulation and experiment research show that the calibration method is effective and the accuracy of the 3-axis mounting angle could reach 10 arcseconds.5.On-ship experimental research for the star sensor.A on-ship experiment was car-ried out in the YuangWang 6 measuremnt ship for the star sensor and the gyro unit.The data obtained on-board was classified into 3 categories,the static,the quasi-static and the dynamic.The data under the quasi-static condition was processed by the commonly used methods for star images under static condition.As a result,the identification result keeps identical with the static conditions.The motion-blurred star image under dynamic condition was processed by the proposed RCRM method,after which the SNR and the identification rate have been improved,and the results are quite close to the ones under static condition.The ACF approach,again,was verified by the on-ship star images.With the ACF approach,the random noise aroused by the motion is reduced and the attitude accuracy is improved.The improvement is approximately proportional to the square root of the number of correlated star image frames.