The antagonistic behavior of the crowd often exacerbates the seriousness of the situation in sudden riots, where the spreading of antagonistic emotion and behavioral decision making in the crowd play very important roles. However, the mechanism of complex emotion influencing decision making, especially in the environment of sudden confrontation, has not yet been explored clearly. In this paper, we propose one new antagonistic crowd simulation model by combing emotional contagion and deep reinforcement learning (ACSED). Firstly, we build a group emotional contagion model based on the improved SIS contagion disease model, and estimate the emotional state of the group at each time step during the simulation. Then, the tendency of group antagonistic behavior is modeled based on Deep Q Network (DQN), where the agent can learn the combat behavior autonomously, and leverages the mean field theory to quickly calculate the influence of other surrounding individuals on the central one. Finally, the rationality of the predicted behaviors by the DQN is further analyzed in combination with group emotion, and the final combat behavior of the agent is determined. The method proposed in this paper is verified through several different settings of experiments. The results prove that emotions have a vital impact on the group combat, and positive emotional states are more conducive to combat. Moreover, by comparing the simulation results with real scenes, the feasibility of the method is further verified, which can provide good reference for formulating battle plans and improving the winning rate of righteous groups battles in a variety of situations.