光压摄动计算中地影间断的处理

人卫精密定轨中受摄星历（或称精密星历,即状态转移）,可由分析解或数值解提供,相应的定轨方法亦有分析法定轨与数值法定轨之称。对于后者,在一般情况下,现有的常微分方程数值解法（或称积分器）已能满足精度要求,除长弧定轨外,有一定问题是值得注意的,即地影“间断”问题的处理,这关系到如何在保证星历精度的前提下提高计算效率的问题。本文针对这一问题,给出了相应的改进算法,并通过数值验证表明算法的有效性。     

状态转移矩阵的差分算法及其应用

指出用数值积分方法计算状态转移矩阵在程序实现时存在的困难。根据精密定轨和参数解算的实际需要,提出用差分算法,即通过两条接近的轨道的差来计算状态转移矩阵差分算法的优点是程序具有良好的结构且编程简单,其不足之处是差分时可能损失精度。将差分算法与数值积分方法的结果进行比较,提出克服其不足处的方法。     

星载GPS自主定轨的长期稳定性

通过模拟长达100天的星载GPS伪距观测资料,进行卡尔曼滤波定轨仿真计算．重点研究：1)采用简化的动力学模型与简化状态转移矩阵,是否保证滤波的长期稳定性；2)模型误差矩阵Q的选取对滤波定轨精度的影响；3)与事后最小二乘批处理相比较,在简化模型下自主定轨的精度．同时给出了相应的结论．     

星载GPS卡尔曼滤波定轨算法

利用卡尔曼滤波定轨算法，处理神舟4号星载GPS伪距实测资料，重点在于研究卡尔曼滤波中模型误差方差矩阵的选取准则，GPS信号中断或连续野值对递推滤波的影响，如何自主监控滤波的运行状态，即是正常还是趋于发散，目的在于评价该算法用于星上自主定轨长期平稳运行的可靠性．     

天文动力学方程数值积分中的一种有效变步法

利用积分曲线的曲率控制步长的技巧，使天文动力学方程数值解法的精度和速度有较大提高，这种方法适用于天体精密定轨以及一些精度要求高的常微分方程初值问题的数值积分。     

Adams—Cowell方法与KSG积分器的比较

在人造地球卫星精密定轨中,有摄星历等量的计算常采用Adams-Cowell方法,美国Texas大学空间研究中心(CSR)的定轨软件中则采用了一种有别于Adams-Cowell方法的KSG积分器。本文对这两种线性多步法作了全面比较,并用典型算例作了数值验证,列出了两种方法中卫星轨道沿迹误差的状况,以此表明为什么人们常采用Adams-Cowell方法。     

人卫长弧定轨中的摄动计算问题

人卫精度定轨中摄动星历表的计算，常采用数值方法，对于长弧定轨，将会遇到数值稳定问题，有人提出采用Ｅｎｃｋｅ特别摄动法，但要保持数值稳定，关键在于坐标摄动法数值计算中参考轨道的选取，而不是用不用Ｅｎｃｋｅ变换，更有效的方法是一般数值方法中采用稳定化措施。     

采用重置参数的轨道改进算法

当使用精度差的初始根数作定轨计算时,被估值的模型参数会吸收初值中所含误差而偏离其合理数值(如CD约为2.2),使定轨计算过程的RMS已不再变化,但轨道收敛到与实际状态有偏离的轨道上。文中给出的算例采用重置被歪曲的估值模型参数方法,首先以TLE根数为初值用精密定轨程序解条件方程,然后以第一轮迭代计算结果作为初始根数并重置模型参数,再进行第二轮迭代计算,使定轨计算结果收敛到正确轨道上,文中还使用另一颗激光卫星的双行根数作初值验证了该方法的有效性。较好地解决了因初值不准所引起的定轨计算不收敛,或收敛到与实际状态有偏离的轨道上的问题。最终得出的RMS达到厘米级精度。文中图示了两次定轨计算的RMS变化曲线图、残差分布图,迭代过程的资料采用率及定轨计算结果。     

两种观测技术综合精密定轨的探讨

利用T／P卫星的SLR和DORIS实测资料,对两种观测技术综合精密定轨中的加权及其对定轨影响的问题作了初步的探讨。根据所提出的一种经验性的加权方法进行了综合定轨计算,结果表明：对于两种不同技术的观测,相对权选取的恰当与否将影响综合定轨的精度;综合定轨的最优加权不仅依赖于观测资料的精度,还与观测资料的多少和几何分布有关;通过选用最优加权,可使得综合轨的精度优于仅用其中一种技术的定轨精度,综合定轨能有效地提高定轨精度。     

双星定位系统的近地卫星联合定轨同质异质观测数据加权方法及应用

针对双星定位系统的近地卫星联合定轨中的多源观测数据的融合处理问题,建立了同质观测数据的二步系统误差修正的改进的方差分量估计最优加权方法;分析指出异质观测数据的多源融合测量模型本质为多结构多参数的非线性回归模型,建立了异质观测数据的模型结构特征分析和方差分量估计相结合的最优加权方法.设计了两类观测数据最优加权及联合定轨参数估计的实现算法,并以双星及备份星的距离和同质观测数据以及双星距离和与星敏感器测角的异质观测数据为例,进行了联合定轨仿真实验.理论分析和仿真计算结果表明;对于同质观测数据联合定轨,采用二步系统误差修正的方差分量估计法,可以获得比传统的经验加权算法更优的定轨精度;对于异质观测数据联合定轨,通过引入表征模型结构特征的加权因子,与平均加权方式相比,近地卫星及静地卫星的联合定轨精度均得到一定程度的改善.     

An intermediate matching technique for solving two point boundary value problems using the perturbation method

The perturbation method, a numerical method for solving two point boundary value problems (TPBVP), is modified to attempt to improve inherent instability and sensitivity problems associated with the method. The desired solution to the TPBVP is divided into two time intervals. The differential equations required to define a solution to the two point boundary value problem are integrated independently over these shorter segments rather than consecutively over the entire trajectory. The independent integration of the differential equations over approximately half of the trajectory instead of the entire trajectory substantially decreases sensitivity and stability properties associated with the numerical integration. The equations for both time segments can be integrated simultaneously. By this procedure, a system of twice the dimension of the original problem is integrated for a period of time equal to half of the time interval for the original problem. To show the effectiveness of the method, two impulse trajectories which minimize the total velocity increment required to transfer a spacecraft from an Earth orbit into a lunar orbit are calculated.     

A second-order solution to the two-point boundary value problem for rendezvous in eccentric orbits

A new second-order solution to the two-point boundary value problem for relative motion about orbital rendezvous in one orbit period is proposed. First, nonlinear differential equations to describe the relative motion between a chaser and a target are presented considering the second-order terms in the gravity. Then, by regarding the second-order terms as external accelerations, we establish second-order state transition equations. Moreover, the J2 perturbations effects can also be considered in the state transition equations. Last, the initial relative velocity to fulfill a rendezvous is determined by solving the state transition equations. Numerical simulations show that the new second-order state transition equations are accurate. The second-order solution to the two-point boundary value problem on eccentric orbits is valid even if the relative range is farther than 500 km.     

Nonlinear analytic solution for perturbed relative motion using differential equinoctial elements

This paper develops a nonlinear analytic solution for satellite relative motion in J₂-perturbed elliptic orbits by using the geometric method that can avoid directly solving the complex differential equations. The differential equinoctial elements (DEEs) are used to remove any singularities for zero-eccentricity or zero-inclination orbits. Based on the relationship between the relative states and the DEEs, state transition tensors (STTs) for transforming the osculating DEEs and propagating the mean DEEs have been derived. The formulation of these STTs has been split into a set of vector and matrix operations, which avoids directly expanding the complex second-order terms, and thus, the obtained STTs could be easy-to-understand and easy-to-code. Numerical results show that the proposed nonlinear solution is valid for zero-eccentricity and zero-inclination reference orbit and is more accurate than the previous linear or nonlinear methods for the long-term prediction of satellite relative motion.     

Satellite Relative Motion Using Differential Equinoctial Elements

For precise control, to minimize the fuel consumption, and to maximize the lifetime of satellite formations a precise analytic solution is needed for the relative motion of satellites. Based on the relationship between the relative states and the differential orbital elements, the state transition matrix for the linearized relative motion that includes the effects due to the reference orbit eccentricity and the gravitational perturbations is derived. This method is called the Geometric Method. To avoid any singularities at zero eccentricity and zero inclination, equinoctial variables are used to derive the relative motion state transition matrices for both mean and osculating elements. This approach can be extended easily to include other perturbing forces.     

Application of the integral variation method to satellite orbit prediction

The Integral Variation (IV) method is a technique to generate an approximate solution to initial value problems involving systems of first-order ordinary differential equations. The technique makes use of generalized Fourier expansions in terms of shifted orthogonal polynomials. The IV method is briefly described and then applied to the problem of near Earth satellite orbit prediction. In particular, we will solve the Lagrange planetary equations including the first three zonal harmonics and drag. This is a highly nonlinear system of six coupled first-order differential equations. Comparison with direct numerical integration shows that the IV method indeed provides accurate analytical approximations to the orbit prediction problem.Advanced Systems Studies; Bldg. 254EElectro-Optical Systems Laboratory; Bldg. 201.     

Studies in the application of recurrence relations to special perturbation methods

The integration by recurrent power series of certain differential equations occurring in celestial mechanics is shown to be very much more efficient and accurate than that produced by classical one step methods. It is shown that for any such system of differential equations the machine time taken to carry out an integration is a minimum for a certain choice of the number of terms taken in the recurrent power series. In the two-body orbits considered this number is about 15. For the same accuracy criterion the power series is faster than the Runge-Kutta method of the fourth order by a factor which varies between 6 and 15 depending on the eccentricity of the orbit.

     

Stabilization of finite difference methods of numerical integration

To examine the stabilizing effects of a modification of the classical finite difference methods of numerical integration the differential equations of perturbed Keplerian motion are integrated for two examples: an artificial satellite of the Earth, and Hill's variation orbit. The modified methods remove much of the instability that is inherent to the classical methods.Presented at the Conference on Celestial Mechanics.