Abstract: In response to the need for terminal speed control of hypersonic aircraft, the dimensionless overload characteristics of the guidance law were analyzed firstly on the basis of the optimal guidance law for landing angle constraints. The analysis results show that the guidance parameter adjustment can impact on the terminal speed. And then a terminal speed control strategy was proposed for the guidance law, and based on this, an online guidance law design strategy was proposed to adapt aerodynamic deviation. On this basis, a long short-term memory (LSTM) deep neural network algorithm was used to carry out online design algorithm of the guidance law parameters. Finally, a trajectory simulation was carried out to verify the online design algorithm of designed guidance parameter. The simulation results show that under the premise of ensuring the terminal accuracy and landing angle within the index requirements, the proposed method can significantly improve the ability of terminal speed control.