Research On Star Sensor-based Space Debris Detection And Postioning Technology

With the increasing number of spacecrafts working in earth orbits year by year,space environment and space security situation have gradually deteriorated.As an important part of space situational awareness,space-based space debris surveillance is of great significance for understanding space situation and protecting space resources.Accurate detection and location technology of space dim and small targets is the technical basis of space-based space debris monitoring.Several difficulties in space dim and small target detection and location technology are as follows: 1.Due to its weak energy,the number of pixels occupied by the target is small,and it does not have the characteristics of shape and texture.Moreover,its imaging is similar to a large number of background stars,which is easy to form a large number of false alarm targets.Therefore,compared with other large targets,the difficulty of recognition and tracking is greatly increased.2.The detected space dim and small targets need to be accurately located,including the right ascension and declination data of moving targets.However,the positioning accuracy of the existing space-based monitoring platform is limited,which affects the high-precision measurement ability of the position of high orbit dim and small targets.Star sensor is one type of high-precision space attitude measurement device,With the continuous development of star sensor technology,the domestic ultra-high-precision star sensor cannot only detect stars with the limit magnitude of 12 currently,but also shoot dim and small targets in space in the field of view.By applying the performance advantages of star sensor to space target monitoring,the principles of star point extraction and star map recognition of star sensor are used to distinguish the dim and small targets from stars.The rapid and effective recognition and high-precision positioning of space dim and small targets can be realized in order to provide ideas for the expansion and application of star sensor.This thesis takes a national advanced research project as the background to carry out the research.For the monitoring task of high orbit stationary point targets,the technical strategy of combining star map recognition based on star sensor and Kalman filter is adopted to realize the recognition and tracking of dim and small targets at high orbit stationary points in the whole celestial sphere.The main content and innovative points are summarized as follows:1.According to the characteristics of star images,the characteristics of weak and small targets,the relationship between the energy of star targets and the magnitude of stars and the observability of space debris have been analyzed.Then,the feasibility of space debris detection and location based on star sensor is analyzed.On the basis of basic concept of star catalog,the selection of the appropriate star catalog in the star image recognition algorithm of weak and small targets based on star sensors is introduced.Meanwhile,the positioning accuracy of various centroid positioning methods is compared.The experimental results show that the Gaussian surface fitting method has the highest accuracy and is the most stable.Under the simulation condition of noise standard deviation of 2,the positioning accuracy of Gaussian surface fitting method can reach 0.02 pixel.2.Aiming at the problem of all celestial sphere star map recognition such as small field of view and high magnitude,an all celestial sphere autonomous star map recognition algorithm is proposed based on improved grid pattern.The algorithm mainly includes two steps: initial matching step and verification step.Firstly,the improved grid pattern is used as the matching feature to determine the candidate matching navigation star of the observation star to be identified,and then the field of view constrained voting is used to screen the initial recognition results and output the final matching results.The introduction of multiple collimators greatly improves the reliability of grid mode construction.The experimental results show that the algorithm can successfully recognize the all celestial sphere star map with the limit magnitude of 10.5Mv.Compared with the grid algorithm,the proposed algorithm has stronger anti-interference ability to position noise and brightness noise.The specific performance is as follows: when the position noise is 2.0 pixel,the recognition success rate of the proposed algorithm is 94.3%,which is 2.9% higher than that of the grid algorithm;When the luminance noise is 0.3m V,the recognition success rate of the proposed algorithm is 96.4%,which is 4.1% higher than that of the grid algorithm.3.Aiming at the problem of weak and small target detection under the background of star images,a target detection technology path is proposed based on local sky reference star image comparison.The proposed algorithm can generate a reference star image quickly through the attitude data,and complete the distinguishment between star targets and non-star targets in the images through the neighborhood matching principle in order to realize the detection and recognition of targets in the star background.The experimental results show that the proposed algorithm can effectively complete the task of detecting weak and small targets in the star background.When the position noise is 2.0 pixel and the brightness noise is0.3Mv,the target recognition success rate can reach 97.6%.4.Aiming at the problem of weak and small target positioning in the star background,combined with the initial state estimation of the target established in the star image recognition stage,the target tracking and positioning can be realized through window scanning by the target tracking algorithm of Kalman filter.The experimental results show that the proposed algorithm can effectively complete the target recognition and tracking positioning tasks with Gaussian noise standard deviation of 5,10 and 15,respectively,and the positioning accuracy of the algorithm for the target can reach 1.22 ",2.15" and 3.30 ",respectively.5.The modules such as star point centroid extraction,all celestial sphere star map recognition,attitude calculation,local sky star map recognition and Kalman filter are transplanted on the embedded platform.The algorithm of each functional module suitable for the operation of the embedded platform is selected.The corresponding module program is written,debugged and compiled.The program runs successfully in the embedded system.The experimental results show that the three-axis attitude calculation accuracy of the star sensor based on ZYNQ platform reaches 0.19 ",0.05"and 3.76 "respectively under the ideal star map test conditions.The average running time of star map data from star point extraction to target position calculation is 26 ms.