Trust Region Strategy for Damping Oscillation in Sequential Convex Programming for Trajectory Optimaization

In response to the demands of online trajectory planning for hypersonic vehicles, this paper introduces a trust region update strategy designed to ameliorate the non-convergence issue resulting from oscillation phenomena, which currently presents in Sequential Convex Programming (SCP). Initially, the non-convex original problem of hypersonic vehicle trajectory planning is transformed into a convex sub problem, using methods such as sequential convexification, relaxation variables, and penalty functions. Subsequently, to resolve the non-convergence challenges arising from oscillatory phenomena, the Damping Oscillation Trust Region Strategy (DOTRS) is devised. Ultimately, the strategy is validated to effectively circumvent the oscillation phenomenon that slows down the iterative process in later stages. Otherwise, a comparative analysis with the Sequential Quadratic Programming (SQP) method demonstrates that the proposed algorithm meets the required solution accuracy, exhibiting faster solution acquisition and superior effective properties.