全天时惯性/天文深组合星图增强及误差抑制技术研究

The application of star tracker is limited by the noise of the sky background radiation in the daytime when used in near earth zone. This project researches on INS/CNS deeply-coupled integration technology in order to resolve this problem. An attitude correlated-images adding (ACIA) method, by which the dynamical star images can be adding together and the signal to noise rate (SNR) will be improved by correlation transformation, is proposed, so the star tracker is able to overcome the limitation of SNR for a single frame, and detect the stellar signal in the daytime. A deeply-coupled integration algorithm for inertial navigation system (INS) and celestial navigation system(CNS) based on the observation of the star image location prediction errors is proposed, by which the systemic errors can be estimated. The star tracker attitude calculation is unnecessary in this algorithm, so the Kalman filter can work even under the condition that there are not enough stars in the field of view, and the robustness of the algorithm is improved. The INS/CNS deeply-coupled integration technology can fully take the advantages of the INS and CNS, reduce the errors of both, and achieve the optimal estimation of systemic errors. The ability of stellar signal detection and the feasibility of deeply-coupled integration algorithm is validated by simulations and dynamic experiments. Finally, the all-day stellar signal detection ability and the high-accuracy and automatically navigation of the INS/CNS integrated system can be achieved.

为突破星敏感器在近地面、白天条件下使用时易受强背景噪声影响的限制，本项目围绕惯性/天文深组合技术开展研究。提出一种姿态关联图像叠加算法，通过关联变换解决动态星图的叠加和信噪比增强问题，突破单帧星图的星光探测极限，实现白昼星光探测；提出一种基于星点预测误差观测的惯性/天文深组合导航算法，通过对星点预测误差的观测实现系统误差的最优估计，避免了星敏感器姿态求解过程，即使视场内可用的星点数较少的条件下也能使滤波器正常工作，提高算法的可靠性；通过惯性/天文深组合算法深度挖掘惯性导航系统和天文导航系统的精度优势，实现两者之间的互相误差抑制，最终实现综合误差的最优估计和最佳组合。通过仿真和动态试验验证星敏感器的全天时星光探测能力及深组合算法有效性，最终实现惯性/天文组合导航的全天时星光探测和高精度自主导航。

星敏感器在近地使用时，受白天强烈的背景光影响，星光探测信噪比急剧降低，暗弱天文目标探测难度极大，且可见星数较少，数据有效率较低，难以实现与惯导的最优组合。本项目针对惯性/天文组合导航系统中全天时星光探测和信息深度融合两大难题展开研究。首先通过理论计算与仿真，分析星光探测的信噪比，为全天时星敏感器的光学系统设计提供理论依据；为了改善星图的非均匀响应特性，提出了一种改进型非均匀校正算法；提出了一种姿态关联图像叠加算法，通过关联变换解决动态星图的叠加和信噪比增强问题，突破单帧星图的星光探测极限，实现白昼星光探测；提出一种基于星点预测误差观测的惯性/天文深组合导航算法，通过对星点预测误差的观测实现系统误差的最优估计，避免了星敏感器姿态求解过程，即使视场内可用的星点数较少的条件下也能使滤波器正常工作，提高算法的可靠性；借助项目组研制的捷联惯导系统和红外星敏感器，开展算法性能的地面惯性验证实验。仿真及实验结果表明：本项目提出的关联叠加算法可以有效地增强图像信噪比，从而增加视场内可见的恒星数量，同时提高星点的质心定位精度；在小视场条件下，相对于传统的惯性/天文松组合算法，本项目提出的深组合算法的位置、姿态精度更高，且可以同时实现系统的全部误差参数的估计和补偿。