期刊:IEEE International Conference on Data Science in Cyberspace [Institute of Electrical and Electronics Engineers] 日期:2024-08-23卷期号:: 308-313被引量:1
标识
DOI:10.1109/dsc63484.2024.00048
摘要
This paper introduces an innovative approach to addressing the Weapon Target Assignment (WTA) problem, utilizing the Deep Q-Learning (DQN) algorithm. The goal is to optimize the allocation between weapons and targets to achieve the most effective combat outcome. Initially proposed by Manne, the WTA problem is a multi-objective, multi-constraint NP-complete problem, the complexity of which increases with the number and variety of weapons and targets. The DQN model presented in this paper transforms the WTA problem into a reinforcement learning issue, effectively addressing the issues of low efficiency and the propensity to fall into local optima associated with traditional heuristic algorithms in large-scale scenarios.