System Vulnerability Oriented to Strategies

System vulnerability has received more and more attention. However, traditional methods of vulnerability are difficult to apply to multi-agent systems with complex game relations with constraints, and research on competition and cooperative relationships rarely considers its system background and constraints. This paper summarizes the existing research and uses reinforcement learning technology to abstract the complex relationship with constraints into a multi-agent model. Through simulation analysis, the impact of three different strategies on system vulnerability is obtained, and provides solutions to improve the vulnerability of the system.