Abstract: To solve the problems of large overload demand and wide distribution of impact speed caused by the consideration of both impact angle and impact speed constraints in the terminal guidance stage of a missile, a terminal guidance law based on virtual expected impact angle is proposed. Firstly, the concept of virtual expected impact angle is defined, and the transition function is designed to reduce initial the overload demand of terminal guidance. Then, the influence of transition function parameters on the impact angle and impact speed is analyzed, and the predictor corrector method is designed to calculate the expected transition function parameters. In order to improve the prediction efficiency and accuracy, the deep neural network is used to train the trajectory data set offline. In actual flight, the extended Kaman filter algorithm is used to identify aerodynamic parameter perturbation online to improve the adaptability. Monte Carlo simulation results show that the proposed algorithm can reduce the overload requirement at the initial stage of terminal guidance, and the impact speed accuracy is within ±15 m/s on the premise of satisfying the impact angle constraint and position accuracy.

Key words: Virtual expected impact angle, Predictor corrector method, Deep neural network, Aerodynamic parameter identification