Reinforcement Learning-Based Intelligent Decision-Making for Infrared Decoy Deployment

Infrared decoys are a critical terminal defense measure for airborne platforms against infrared-guided short-range air-to-air missiles, where the quality of the interference strategy largely determines whether an aircraft can successfully evade incoming threats. To overcome the limited adaptability of traditional rule-based strategies in complex threat scenarios, this paper proposes an intelligent decision-making method for infrared decoy countermeasures based on reinforcement learning. First, a simulation environment is constructed that integrates missile dynamics, aircraft maneuvers, infrared decoy motion, and centroid interference models. The interference decision problem is then formulated as a Markov decision process, with carefully designed state and action spaces as well as a tailored reward function. Finally, a Proximal Policy Optimization (PPO) algorithm is employed to train the decision-making agent, enabling it to autonomously learn optimal countermeasure strategies under various threat conditions. Simulation results demonstrate that, compared with conventional rule-based methods, the proposed approach increases the aircraft survival probability from 58% to 92% while satisfying the real-time requirements of terminal defense. This method provides an efficient and intelligent decision-support framework for airborne platforms facing infrared-guided missile threats.