Real Time Solutions to Aerospace Optimal Control Problems

Abstract

A real-time autonomous trajectory generation methodology for aerospace vehicles using pseudo-spectral based global collocation methods is presented in this thesis. The proposed method is formulated in a manner whereby it can converge from random initial guess pro les in real time. The proposed methodology is validated using Monte-Carlo simulations for various practical aerospace problems to demonstrate the robustness in presence of initial condition errors. In the thesis, the solution to the trajectory optimization using both Direct and In- direct methods are presented. A two level optimization scheme is adopted for solving the trajectory and minimizing the fuel consumption using the Direct method . In the outer level, the required burn duration for achieving mission objectives is iterated using a simple optimization procedure for minimizing fuel consumption. In the inner level, the trajectory for a given burn time obtained from the outer level, is solved using a modi ed Newton-Raphson scheme. An important advantage of this scheme is that, it does not use any general purpose, commercially available gradient-based optimization routines as they are not suitable for real-time implementation. In the second part of the thesis, the proposed algorithm is applied to general optimal control problems involving states and co-states. In this scheme, the optimal solution for the states, control and co-states are computed using a discretized version of the Indirect method. The proposed method can handle inequality constraints on states and control as well. The effectiveness of the algorithm is demonstrated for a set of general optimal control problems with emphasis on aerospace trajectory optimization