Finite-Time Convergent Missile Terminal Guidance Law Based on Deep Neural Network

This paper proposes a finite-time convergence guidance law based on deep neural networks (DNN-FTG). The guidance law uses a large amount of simulation data from the finite time input to the state stabilization (ISS) guidance law to train the deep neural network. Then, DNN-FTG is simulated and compared with the finite-time ISS guidance law, and the guidance performance of DNN-FTG is analyzed by miss distance and energy consumption. The neural network's performance is verified by giving different initial conditions, maneuvering forms, and anti-interference tests. The results show that DNN-FTG can replace the finite-time ISS guidance law for real-time missile control, and DNN-FTG has better guidance performance, generalization, and anti-interference.