Hp-Adaptive RPD based sequential convex programming for reentry trajectory optimization

Abstract

Sequential convex programming (SCP) methods have been developed to solve reentry trajectory optimization problems. Due to the oversimplified discretization and iteration, the accuracy and efficiency of the existing SCP methods can be further improved. In this paper, a SCP algorithm based on the hp-adaptive Radau pseudospectral discretization (RPD) is proposed. In the proposed algorithm, the iteration process is divided into three stages depending on the characteristics of subproblems. The constraint relaxation technique is applied in the first stage to ensure that the iteration is stable. During the second stage, the number and position of discretized points will be updated adaptively according to the discretization error and the curvature of state. In the last stage, the linearization error is reduced by several iterations without updating mesh, and the regularization technique is utilized to improve the convergence rate of this process. The proposed algorithm is validated and examined by a typical reentry example. With comparable or even higher results accuracy, the CPU time reduced by 40%-70% when compared to other SCP methods, and is only twentieth of that of GPOPS-II.