高精度导星测量系统细分定位技术研究

	高精度导星测量系统细分定位技术研究
	刘南南
学位类型	博士
导师	徐抒岩
	2014-07
学位授予单位	中国科学院大学
学位专业	光学工程
摘要	高精度导星测量系统是大口径空间天文望远镜稳像控制系统的重要组成部分，是最精密的角位置测量元件。它利用主光学系统超长的焦距极其相关算法获得惯性角位置偏移信息，以图像的方式为精级稳像控制系统提供反馈信息。天文望远镜主光学系统焦距确定的情况下，星点细分定位精度直接决定了导星测量系统的惯性角位置测量精度。本文以此为研究背景，主要研究适用于导星测量系统的细分定位技术，从软件算法的角度提高测量精度，实现精级稳像控制系统的闭环控制。提出了基于最小二乘支持向量机的系统误差校正算法，解决了CMOS探测器填充因子低，感光区形状影响细分定位精度的问题。分析研究统计学习理论，详细介绍最小二乘支持向量机回归数学模型，以此为理论基础建立星点定位系统误差校正的模型，利用蒙特卡罗数值仿真的方法，用带有高斯径向基函数核的最小二乘支持向量机进行回归分析，得到系统误差与星点质心的理想位置和感光区形状大小的非线性函数关系，并用此函数关系对质心估计值进行后校正。提出了基于自适应Kalman滤波的亚像元细分算法，解决了高动态条件下的导星跟踪问题并发挥CMOS探测器可任意开窗的特点。研究了Kalman滤波原理，提出了采用预测开窗和Kalman滤波相结合的星点定位方法。预测开窗方法利用陀螺输出的惯导信息得到星点的粗位置，星点的量测位置由星点预测位置较小的窗口范围内确定，用Kalman滤波器对粗位置滤波，输出高精度的星点位置。由于星点量测位置在小窗口内计算，提出的算法运算速度快，同时Kalman滤波器能有效的抑制随机噪声，也特别适合于导星测量系统这种高星等低信噪比星点图像，能提供高精度的星点位置信息。提出了迭代加权质心算法，解决了高星等低信噪比图像的细分定位及像移和像差使光斑椭圆化的问题。运用统计信号处理理论首先对光斑信号建模，推导出光斑中心的最大似然估计公式，由于最大似然估计是一个性能良好的无偏估计子，它构成了的理论基础；给出了采用迭代加权质心法的计算流程，用统计信号处理理论中Cramer-Rao下限推导出当迭代次数无穷大时，星点定位的极限精度。迭代加权质心法在算法收敛时，本质上是用一个高斯权函数对光强加权，增加了图像的信噪比，而且在两个方向可用不同的高斯宽度，适用于椭圆光斑，通过数值仿真实验验证了提出的算法具有良好的性能。　　搭建了导星测量地面实验系统，对提出的算法进行验证。介绍实验系统的组成和工作原理；对实验中相关器件的参数进行分析；分别通过系统误差补偿实验、静态精度测量实验和动态精度测量实验对提出的算法进行实验验证，取得了预期的实验效果。　　实验结果表明，在导星测量系统工作条件下，提出的算法的测量精度相比于传统的细分定位算法有很大提高。研究成果对我国空间天文望远镜高精度导星测量系统的研制具有一定的借鉴意义。
其他摘要	Fine guidance sensor is a high-precision measurement of large-aperture space telescope image stabilization important part of the control system , is the most precise angular position of the measuring element. It uses the main optical system is extremely long focal length correlation algorithm to obtain highly accurate inertial angular position offset information in a way to fine -level image stabilization control system provides feedback information to achieve fine control level stabilization loop control system . In the case of the primary optical system, the focal length of the telescope determined , star point positioning accuracy subdivision directly determine the angular position measurement accuracy inertial measurement guide star system . In this paper as a background , the main research guide star measurement system suitable for subdivision positioning technology . 　　CMOS detectors for low fill factor, affecting the shape of the photosensitive area subdivision positioning accuracy is proposed based on least squares support vector machine algorithm for error correction system . Analysis of statistical learning theory , detailing least squares support vector machine regression mathematical model , then as a theoretical basis for the establishment of the star point positioning system error correction model , and finally, the numerical simulation using Monte Carlo method, with Gaussian radial basis function kernel least squares support vector machine regression analysis, the non-linear function of system errors and centroid location and the ideal size and shape of the light-sensitive area , and use this as a function of the heart after a confrontation estimates corrected.Tracking for high dynamic conditions and guide star problems CMOS detector can play any windows features , is proposed based on adaptive Kalman filtering sub-pixel subdivision algorithm . First study the Kalman filter theory, the star point positioning method is proposed to predict window and using a combination of Kalman filtering . Coarse location prediction method uses the gyro output window INS information is star point star point in the measurement position predicted by the smaller star point range to determine the position of the window , and then , using the Kalman filter for filtering coarse location , high output Star point accuracy . As the star point measurement position calculation in a small window , while the Kalman filter can effectively suppress random noise, is particularly suitable for measuring guide star system of such high magnitude star point low SNR image , can provide highly accurate satellite position information . 　　Low -noise ratio for high- magnitude image and the image shift and errands like oval spot problems , the use of statistical signal processing theory first spot signal modeling to derive maximum likelihood estimation formula spot center.Then proposed an iterative weighted centroid calculation processes, statistical signal processing theory to derive the Cramer-Rao lower limit when the number of iterations to infinity, the star point positioning accuracy limits . When iterative weighted centroid algorithm converges , through numerical simulation results show that the proposed algorithm has good performance. 　　Build a ground test measuring guide star system, the proposed algorithm for authentication. First introduced the composition and working principle experiment system ; experiment parameters related devices for analysis ; compensated by the system error experiments , static and dynamic accuracy of the measurement accuracy of the experimental measurement experiments conducted experiments to verify the proposed algorithm achieved the expected experimental results . Experimental results show that, under operating conditions guide star measurement system , the measurement accuracy of the proposed method compared to the conventional positioning subdivision algorithm greatly improved. Research results have certain significance for the development of China's space telescope guide star precision measurement system .
语种	中文
文献类型	学位论文
条目标识符	http://ir.ciomp.ac.cn/handle/181722/41438
专题	中科院长春光机所知识产出
推荐引用方式 GB/T 7714	刘南南. 高精度导星测量系统细分定位技术研究[D]. 中国科学院大学,2014.

条目包含的文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
刘南南.pdf（3135KB）			开放获取	CC BY-NC-SA	请求全文