Abstract: Focusing on stochastic and concurrent task scenarios, the study on modeling and optimal control method of a kill web is carried out. Firstly, a formalized modeling of a kill web based on stochastic controlled Petri net is proposed, which depicts the operational logic of a kill web under stochastic and concurrent strike task and dynamic variation process of a kill web state under different kill chain selection actions. Further, for the solution problem of the optimal kill chain selection strategy, an optimal strategy solution method based on the optimal strategy construction paradigm of Markov decision-making process and deep reinforcement learning is proposed. The research result of this paper can provide theoretical model basis for the modeling and analysis of combat system under complex confrontation scenarios and the relative work of optimal control.