RLV Reentry Trajectory Optimization through Hybridization of an Improved GA and a SQP Algorithm

摘要

A hybrid optimization method combining an improved genetic algorithm with sequential quadratic programming is proposed for the optimum design of the reentry trajectory of a reusable launch vehicle. The advantages of the genetic algorithm of insensitivity to the initial values and global convergence and the advantages of sequential quadratic programming of rapid convergence and high precision were obtained. The weaknesses of the genetic algorithm, including oscillation of the solution, and the weaknesses of sequential quadratic programming, including a small convergence radius, sensitivity to the initial values, and ease of falling into a local extremum, were overcome. An improved genetic algorithm with a simulated-annealing penalty function was employed to search the design space globally, sequential quadratic programming was used for local optimization, and direct collocation was used to discretize the optimal-control problem into a nonlinear programming problem. A global high-precision solution could be obtained without an initial guess because of the reduced sensitivity to the initial values. The results show the correctness, effectiveness, and robustness of the algorithm.