Online Suboptimal Ascent Guidance for Electromagnetic Assisted Launch Vehicle

This paper presents an online suboptimal guidance approach for the ascent phase of an electromagnetic assisted launch vehicle (EMALV). First, the optimal ascent guidance problem for the EMALV is formulated with consideration of the nonlinear dynamics associated with a rotating spherical Earth. Then, the Radau pseudospectral method is utilized to numerically address this problem, and optimal ascent trajectories under various launch conditions and target orbits are obtained. These trajectories are employed to train a neural network, which enables the real-time prediction of guidance commands based on the current states and flight mission. Numerical simulations demonstrate that the ascent guidance commands generated by the proposed method closely approximate the optimal control, and all terminal errors at the orbit insertion point remain small. This validates the strong adaptability and high guidance accuracy of the method. Finally, the effects of launch velocity and launch angle on payload capacity are further analyzed to support the overall design of EMALVs.