Application of direct method transcription for a human-class translunar injection trajectory optimization.

摘要

This thesis presents a new trajectory optimization software package developed in the framework of a low-to-high fidelity three degree-of-freedom (3-DOF)/6-DOF vehicle simulation program named Mission Analysis Simulation Tool in Fortran (MASTIF) and its application to a translunar trajectory optimization problem. The functionality of the developed optimization package is implemented as a new "mode" in generalized settings to make it applicable for a general trajectory optimization problem. In doing so, a direct optimization method using collocation is employed for solving the problem. Trajectory optimization problems in MASTIF are transcribed to a constrained nonlinear programming (NLP) problem and solved with SNOPT, a commercially available NLP solver. A detailed description of the optimization software developed is provided as well as the transcription specifics for the translunar injection (TLI) problem.;This assessment of the final results is formulated via a metric given as the minimization of the TLI main engine burn time, which is equivalent to the maximization of the mass at main engine cutoff (MECO). Key design parameters include the initial values for three orbital angles (right ascension of ascending node, argument of perigee, and true anomaly) and three Euler angles for steering during the main engine burn. To do so, the solution starts by modeling the entire trajectory into three distinct phases. The first two phases are based on a collocation method whereas the third phase appears with a high order Runge-Kutta integration.;The next part of assessing the TLI trajectory utilizes MASTIF's vehicle simulation capabilities (the other "mode" within MASTIF). This includes the ability to design and test new and existing guidance, navigation, and control (GN&C) algorithms. As a demonstration of MASTIF's versatility, results from the trajectory optimization (the open-loop solution) in the form of a set of initial states and specific orbital target parameters at MECO are used in a new preliminary assessment of a variant of the Space Shuttle's flight-proven closed-loop guidance algorithm named Powered Explicit Guidance (PEG). Main engine burn times and the LVLH Euler angles from the open-loop and closed-loop solutions are compared to show approximate agreement and efficacy of MASTIF's two distinct "modes".