Three-Dimensional Two-Stage Impact Time and Angle Control Guidance Strategy Under Time-Varying Velocity

This paper proposes a two-stage guidance strategy for impact time and angle control of vehicles under time-varying velocity. To address the challenge of coupled constraints with nonlinear velocity behavior, a virtual target is introduced to decompose the trajectory into two stages: the first stage uses an optimal angle-constrained guidance law to reach the virtual target, and the second stage employs proportional navigation to hit the real target while maintaining the desired angle. Residual neural networks (ResNet) are adopted to predict the flight time and velocity in the first stage, while the analytical time-to-go formula of time-varying velocity vehicles are derived for the second stage. The flight time is regulated by iteratively adjusting the virtual target position via online Newton–Raphson-based timing calibration methods. Simulation results demonstrate that the proposed method achieves high-precision control of both time and angle constraints.