State-split deep reinforcement learning approach to weapon target assignment problem

In this study, a novel deep reinforcement learning (DRL) algorithm is proposed to deal with the weapon target assignment (WTA) problem, which takes the unmanned aerial vehicle (UAV) safety transition task as the background. Firstly, a reinforcement learning environment is developed, taking into account the heterogeneity of the targets. Secondly, a state split deep Q-learning network (S2-DQN) algorithm is proposed to approximate the optimal solution with a multi-process approach. Finally, simulation and comparison experiments demonstrate the effectiveness and superiority of the S2-DQN algorithm.