High-efficiency reduced space sequential convex programming using low-complexity matrix inversion

In response to the increasing demand for real-time optimization in aerospace systems, reduced space sequential convex programming (rSCP) has been developed to improve the computational efficiency of conventional sequential convex programming by explicitly eliminating state variables and state equati...

Full description

Saved in:
Bibliographic Details
Published in:Aerospace science and technology Vol. 159; p. 109958
Main Authors: Ma, Yangyang, Pan, Binfeng, Huang, Longxin, Chen, Qi, Xu, Zihui
Format: Journal Article
Language:English
Published: Elsevier Masson SAS 01.04.2025
Subjects:
Low-complexity matrix inversion
Reduced space sequential convex programming
Trajectory optimization
Reduced space sequential convex programming
Low-complexity matrix inversion
Trajectory optimization
ISSN:1270-9638
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In response to the increasing demand for real-time optimization in aerospace systems, reduced space sequential convex programming (rSCP) has been developed to improve the computational efficiency of conventional sequential convex programming by explicitly eliminating state variables and state equations from the optimization formulation. However, as the number of discretization nodes increases, this computational benefit diminishes due to the significantly increased computational burden of matrix inversions. To address this challenge, this study proposes three efficiency-enhanced rSCP methods utilizing low-complexity matrix inversion techniques. The first method employs an explicit discretization scheme to achieve closed-form matrix inversions. The second leverages inexact Jacobian information to circumvent the repeated computation of time-consuming matrix inversions in each iteration. The third transforms a single large-scale matrix inversion into multiple smaller-scale ones and their multiplications to reduce computational complexity. Numerical experiments of a minimum-fuel rocket landing problem and a maximum-velocity launch ascent problem are conducted to validate the performance and superiority of the proposed methods. •Efficiency enhanced reduced space sequential convex programming (rSCP) methods with low-complexity matrix inversion.•Proven local optimality equivalence between the proposed rSCP methods and the basic SCP method.•Numerical validation showcasing high efficiency, accuracy, and robustness in rocket landing and launch ascent problems.
ISSN:1270-9638
DOI:10.1016/j.ast.2025.109958