Balancing Global and Local Search via Q-Learning in Evolutionary Algorithms for Air Defense Resource Assignment Problems
针对防空资源分配问题，提出一种利用Q学习动态平衡全局与局部搜索的进化算法，实验表明其优于六种现有算法，能有效提升智能防空系统的协同防御能力。
The coordinated operation of modern air defense systems represents a highly complex engineering challenge, necessitating intelligent decision-making to maximize defensive performance. Central to this framework, air defense resource assignment problems (ADRAPs) require the flexible and coordinated management of radar and missile resources. However, solving ADRAPs is highly challenging due to their sparse decision spaces, complex system constraints, and strict computational requirements. To address these challenges, we formulate a mathematical model tailored to realistic battlefield scenarios and develop a novel evolutionary algorithm that leverages Q-learning to balance global and local search strategies. Specifically, the proposed algorithm incorporates an adaptive mechanism that dynamically selects the optimal local search strategy based on the current population state. Furthermore, we design and integrate a knowledge-guided search method to enhance the efficiency of the local search process. Experimental results across diverse air defense scenarios demonstrate that the proposed algorithm significantly outperforms six state-of-the-art algorithms, effectively improving the coordinated defensive capabilities of intelligent air defense systems.