This dissertation addresses the pressing challenges in space situational awareness (SSA) and space debris management by developing advanced computational methods to enhance the precision and efficiency of orbit propagation and trajectory optimization. The research emphasizes the application of orthogonal polynomials—specifically Chebyshev and Legendre polynomials—known for their superior approximation properties, and the innovative use of inverse-optimal control in astrodynamics.
Chapter 3 of the dissertation introduces several advancements in gravity modeling. First, it introduces a normalized, singularity-free Cartesian formulation of the spherical harmonics model. This formulation overcomes the traditional model's limitations such as pole singularities and computational inefficiency. Additionally, it incorporates a finite element model (FEM) that is locally valid and reduces the computational demands by adapting the degree of spherical harmonics based on the radial distance, particularly beneficial for satellites in elliptical orbits. The adaptive FEM approach, coupled with parallel computing techniques, demonstrates potential speedups of up to 200 times, depending on the implementation specifics, thus addressing high computational costs associated with traditional global models.
Chapter 4 shifts focus to numerical integration, moving beyond conventional methods like explicit Runge-Kutta or multistep methods to explore orthogonal polynomial-based methods—Chebyshev-Picard and Gauss-Legendre collocation. Chebyshev-Picard methods are formulated in a matrix form to facilitate efficient parallel processing, significantly enhancing computational speeds in handling high-order gravity fields. The research bridges these methods with implicit Runge-Kutta methods, establishing their stability and accuracy. Innovations such as feedback-accelerated iterations and adaptive time-stepping further refine the numerical integration process, optimizing efficiency and adaptability in practical applications.
Chapter 5 explores trajectory optimization using a Lyapunov-based controller that simplifies the generation of initial guesses for both direct and indirect optimization methods. By employing an inverse-optimal-control framework, the dissertation identifies and minimizes specific cost functions, leading to the derivation of corresponding costates. These costates are utilized to solve time-optimal, orbit-transfer problems effectively, demonstrating the CCM controller's capability to maintain robust performance, even in the presence of gravity perturbations.