Infrared Air Combat Simulation Model for Deep Reinforcement Learning

Abstract

Aiming at the problem of lacking credible and realistic infrared air combat simulation platform for applying the deep reinforcement learning method, this paper explored the design requirements for the construction of simulation system, built the overall architecture of infrared air combat simulation system, described the structure, principle and working process of the fighter jet, infrared air-to-air missile, point source decoy and environment model. The implementation method of the simulation system was given, and the credibility of the system was verified through attack and defense simulation examples and error analysis of missile anti-jamming probability, which indicated that the simulation system can be used for the training and testing of agents based on deep reinforcement learning in infrared air combat scenarios.