Air-to-air missile shoulder-crossing launching guidance law design method based on deep reinforcement learning

Abstract

The invention discloses a deep reinforcement learning-based air-to-air missile shoulder-crossing launching guidance law design method. The method comprises the following steps of 1, performing normalized dynamic modeling on shoulder-crossing launching; normalizing the model to enable each state quantity to have similar magnitude, so that the weight updating of the neural network can be more stable; 2, in order to adapt to a research normal form of reinforcement learning, modeling the research problem in the step 1 into a Markov decision process; 3, an algorithm network is built, and algorithm parameters are set; the selected deep learning algorithm is a near-end strategy optimization algorithm PPO, and 4, before training reaches a target reward value or the maximum step number, the intelligent agent continuously collects state transition data and rewards according to the PPO algorithm, and continuously iteratively updates parameters of the Actor network and the Critic network. By applying the technical scheme of the invention, the missile can obtain the attack angle guidance law with suboptimum and robustness in a complex aerodynamic environment, the limitation of different maneuvering capabilities of the missile is considered, and the method has practical value in future air combats.

Description

technical field [0001] The invention relates to the field of aircraft control, in particular to a design method of a guidance law for air-to-air missile over-the-shoulder launch based on deep reinforcement learning. Background technique [0002] In modern air combat, as the maneuverability of fighter jets continues to increase, the scene of close combat becomes more and more complex. In order to improve the fighter's combat capability in close combat, the over-the-shoulder launch method that can attack targets in the rear hemisphere has become the focus of research. The air-to-air missile launched over the shoulder can quickly change the flight direction after launch, and enter the terminal guidance stage after the seeker locks the target, so that the missile has omnidirectional attack capability. However, the missile will face complex aerodynamic phenomena such as asymmetric vortex shedding and induced torque during the high angle of attack steering phase, which is a typic...

AI Technical Summary

Problems solved by technology

At present, the commonly used method is to use the pilot to track offline and optimize the ballistic trajectory or constant angle of attack, but in the complex aerodynamic environment and the rapidly changing air combat situation, it is easy to lose the target after the over-the-shoulder launch

Images