This article proposes a low-thrust reconfiguration strategy for formation flying based on the Jordan normal form and evaluates its performance based on the required control acceleration and fuel cost. First, the relative dynamics of reconfiguration is disposed by Jordan decomposition to describe the unstable component, i.e., the drift velocity along the track direction, which is not explicit in eigenstructure. Next, the initial states of the Jordan-reduced dynamics are regarded as orbital invariants without control to characterize a formation configuration, similar to the definition of orbital elements that are invariants of the unperturbed Keplerian orbit. Based on the derived differential form of ISJD with control, the reconfiguration trajectory is parameterized by a functional integral, the existence of which is then proved analytically by the proposed polynomial series method. The specified optimal trajectories are yielded by the practical Radau pseudospectral method numerically. Finally, the effects of reconfiguring positions and relative orbital orientation on the required acceleration and fuel cost are investigated using a technique for order preference based on similarity to the ideal solution.

This paper investigates a multi-satellite scheduling method for observing moving targets based on the hybrid heuristic algorithm and observation strip selection method. Firstly, the constraint satisfy problem model for multi-satellite guided collaboration is established considering multiple constraints. Subsequently, based on the genetic algorithm framework incorporated by congestion-based simulated annealing, the enhanced hybrid genetic simulated annealing algorithm is proposed to assign electronic and imaging satellite missions. The uneven distribution of workload among satellites can be balanced by implementing a multi-layer encoding method and neighborhood generation based on congestion, which effectively reduces conflicts and promotes fast convergence. Additionally, for imaging observation strip selection, the short-term track prediction model for moving targets is developed to determine the potential target area, and the area dynamic decomposition is adopted to further locate the specified imaging strip with the highest discovery probability. Finally, numerical results of both hetero-orbit and co-orbit constellations verify the feasibility and effectiveness of the proposed multi-satellite scheduling method.

This paper proposes a semi-analytical method combining Poincaré Contraction Mapping (PCM) with Conditional Variational Autoencoder (CVAE) for efficient screening of bounded relative orbit across perturbed environments. The PCM projects high-dimensional system parameters to a two-dimensional (2D) feature parameter pair of crossing period and separation angle, which is surjective but non-injective. To address the one-to-many inverse mapping challenge, a CVAE-based deep learning model is developed to raise the dimensionality from the 2D feature parameters back to the state space, enabling rapid and diverse generation of long-duration relative orbits with bounded amplitude. And the PCM-CVAE method is validated in both Earth-centered displaced orbits (DO) and Earth-Moon libration point orbits (LPO), demonstrating consistent generality. Moreover, the accuracy and dispersion of CVAE method are evaluated and compared with ANN and KNN-GMM algorithms. Results of the scenario of DO demonstrate that the time-angle accuracy of PCM-CVAE method realizes an average value (0.53%, 2.48%) and an optimal value (1.92×10^-4, 5.94×^-4). Meanwhile, compared to the previous traversal search method by finding intersection of contour maps, PCM-CVAE reduces the search time from 6.44 hours to 7.83 minutes, achieving 98% reduction in computational cost with high accuracy.

This paper investigates the reconfiguration optimization of the formation flying in elliptical orbits. Based on the Tschauner-Hempel equations, the time-varying system has been reduced to a time-independent one by Lyapunov-Floquet transformation without losing its relative motion characteristics. A geometric configuration invariant of the formation is introduced and general relative configurations are considered as linear combinations of essential components in proportion to this invariant. Benefitting from the evolution of the configuration invariant, the reconfiguration optimization is converted into an optimal parameter selection problem and the transfer trajectory is parameterized by a functional integral. In particular, the indirect method of optimal control based on the reduced dynamics avoids the time-varying derivative of costate variables, which simplifies the optimization problem. The numerical results of both low-thrust and impulsive reconfigurations verify the effectiveness of the reconfiguration optimization based on the reduced relative dynamics.

The existing atmospheric mass density models (AMDM) would produce considerable errors in orbital prediction for Low Earth Orbit (LEO) satellites. In order to reduce these errors and correct the AMDM, this paper presents methods based on data mining with historical data of two-line element (TLE). Starting from a typical LEO satellite, TIANHUI, two orbital dynamical models are firstly proposed as the simulation environment to generate training data. The historical TLE data are regarded as actual space environment and used to generate application data. Secondly, three data mining methods, Random Forest (RF), Artificial Neural Network (ANN) and Support Vector Machine (SVM), are combined with the training data to investigate their feasibility in recovering the known deviation of AMDM under simulation environment. Training results show that RF displays the best performance and achieves the accuracy of 99.99%, while the other two methods only achieve 86.83% and 71.90% respectively. Thirdly, under the actual space environment, this paper uses new training and application data to research the ability of the three methods in recovering the unknown deviation of the AMDM and improve the accuracy of orbital prediction. Numerical results are evidential to the accuracy of the proposed methods based on data mining. It is concluded that the capabilities of the data mining for correction for the atmospheric model are very promising, with great potential to advance practical applications on on-orbit propagation.