Thursday, December 16, 2021, 9:00
Averaging Techniques offer a way to largely reduce dimensionality of the problem and therefore the computation times by modeling the incremental changes of the orbit parameters over one orbital revolution. This allows very large integration steps, e.g. in the order of several days. The faster computation is traded for the ability of bringing the spacecraft to a specfic location on the orbit. Orbit averaging is used to solve a minimum-time problem as a preliminary design step that finds a suitable initial guess for the minimum-propellant problem which is solved subsequently. The NLP resulting from transcription of the optimal control problem using orbital averaging is able to converge from a rather poor initial guess since it removes the oscillatory behavior of the state and control trajectories. The minimum-propellant problem is constructed by direct application of the dynamics, i.e. no averaging technique is used. Unlike the averaged dynamics strong oscillation appear in the state and control trajectories. Therefore a higher quality initial guess is required and the resulting nonlinear problem is much larger than the previous one. It enables the positioning of the spacecraft on the target orbit. This thesis presents a methodology for the optimization of low-thrust transfers with hundreds of revolutions. It combines averaging methods for initialization and direct transcription with continuous integration. Averaging technique is used to provide an initial guess for the NLP constructed by direct transcription which is then solved to satisfy the terminal state constraint. The methodology has been demonstrated by solving a GTO-GEO transfer.
Meeting-ID: 627 9173 7415