A novel robust hybrid gravitational search algorithm for reusable launch vehicle approach and landing trajectory optimization

• The reusable launch vehicle (RLV) approach and landing (A&L) trajectory is formulated into two sub-phases: glide and flare.
• A novel robust hybrid optimization algorithm combined advantages of gravitational search algorithm (GSA) and gauss pseudospectral method (GPM) is proposed.
• The GSA is improved by adding the elite memory reservation strategy and adaptive gravitational constant adaption.
• The hybrid algorithm is firstly applied to the trajectory optimization of RLV A&L, comparative study is presented to further demonstrate its superiority.

The approach and landing (A&L) trajectory optimization is a critical problem for secure flight of reusable launch vehicle (RLV). In this paper, the A&L is divided into two sub-phases, glide phase and flare phase respectively. The flare phase is designed firstly based on the desired touchdown (TD) states. Then, the glide phase is optimized using a proposed novel robust hybrid algorithm that combines advantages of the gravitational search algorithm (GSA) and gauss pseudospectral method (GPM). In the proposed hybrid algorithm, an improved GSA (IGSA) is presented to enhance the convergence speed and the global search ability, by adopting the elite memory reservation strategy and an adaptive gravitational constant adaption with individual optimum fitness feedback. At the beginning stage of search process, an initialization generator is constructed to find an optimum solution with IGSA, due to its strong global search ability and robustness to the initial values. When the change in fitness value satisfies the predefined value, the IGSA is replaced by the GPM to accelerate the search process and to get an accurate optimum solution. Finally, the Monte Carlo simulation results are analyzed in detail, which demonstrate the proposed method is practicable. The comparison with GSA and GPM shows that the hybrid algorithm has better performance in terms of convergence speed, robustness and accuracy for solving the RLV A&L trajectory optimization problem.