Rapid trajectory optimization for hypersonic vehicles based on line search sequential convex programming

Abstract

For the hypersonic vehicle reentry rapid trajectory optimization problem, a sequential convex programming method based on a line search algorithm is proposed. The original nonlinear optimal control problem is transformed into a convex optimization problem through convexification and discretization, and then solved using sequential convex programming. To reduce linearization error, a transformation is applied to decouple path constraint variables. To improve convergence, a golden section line search algorithm is designed and integrated into the sequential convex programming framework. Unlike traditional backtracking line search methods, the golden section strategy avoids the need for gradient computations, thereby relaxing the requirement for differentiable objective functions and improving computational efficiency with a predefined convergence interval. In addition, a terminal constraint compensation mechanism is introduced to address the “pseudo-optimal solutions” issue caused by the line search. Simulation results demonstrate that the proposed method is both effective and fast, achieving solution times more than 10 times shorter than conventional trajectory optimization methods. Comparative experiments further validate the advantages of the golden section line search algorithm in reducing iterations and enhancing convergence speed.