107 / 2024-08-15 20:40:21

Intelligent Decision-Making Method for Non-Orbital Segment Flight Timing under Launch Vehicle Thrust Failure

Launch vehicle,Intelligent decision-making,LSTM,Pseudospectral method

During the non-orbital phase of a launch vehicle's flight, a thrust failure can significantly increase the flight duration, potentially resulting in an orbital insertion time deviation of up to several hundred seconds, complicating subsequent orbital transfer missions. To enhance the computational efficiency of trajectory re-planning during the non-orbital phase under propulsion failure conditions, this paper proposes an intelligent decision-making method for flight timing based on deep learning. By using the pseudospectral method, different thrust reduction failure scenarios are solved offline, and a "failure mode-flight timing" sample set is established. During actual flight, the current failure state, failure time, and flight state are used as inputs to an online Long Short-Term Memory (LSTM) network, which rapidly determines shutdown time and program angle commands, facilitating online trajectory re-planning under propulsion failure conditions. Simulation results demonstrate that the computation time of the intelligent decision-making process meets real-time requirements, making it suitable for online decision-making while effectively predicting various flight trajectories.