Abstract: To address the cooperative jamming problem of networked jammers in long-range support jamming scenarios, this paper proposes a deep reinforcement learning-based jamming resource allocation method. The jamming resource allocation model is formulated as a multi-agent Markov decision process. Furthermore, an improved Multi-Agent Proximal Policy Optimization (H-MAPPO) algorithm is proposed, which introduces a multi-branch Actor network to jointly optimize discrete and continuous parameters, thereby enhancing the convergence speed of hybrid action space training. Simulation results demonstrate that the proposed cooperative jamming method significantly outperforms the baseline algorithm in terms of convergence speed, jamming effectiveness, and resource utilization efficiency, achieving superior jamming effects on networked radars and providing an efficient and stable resource allocation strategy for long-range support cooperative jamming.