Abstract: This paper researches the ascent trajectory optimization model and the optimization method for a reusable launch vehicle, which takes off horizontally, and climbs and accelerates using a multi-combined-cycle-based engine. Firstly, in consideration of the fact that the synergistic work of multiple types of engines is required to accomplish the ascent flight, and the complicated nonlinear coupling among propulsion, aerodynamics, trajectory, and performance index, the mathematical models of the propulsion and aerodynamic coefficients are proposed. Secondly, in order to reduce the nonlinear coupling, a novel ascent flight profile is proposed; in this way, the key optimization parameters are picked out, the trajectory constraints are easily satisfied, and the number of the constraints tackled by the optimization method is reduced. Thirdly, an improved particle swarm optimization (PSO) method is proposed to optimize the trajectory. Based on the analysis of the convergence of PSO, a reinforcement learning mechanism is introduced into PSO to automatically, intelligently, and adaptively control the searching process of PSO. The numerical simulation indicates the efficiency of the proposed method.

Key words: Combined cycle engine, Trajectory optimization, Particle swarm optimization, Reinforcement learning