Abstract: With the wide application of artificial intelligence technologies such as deep reinforcement learning in real-time strategy games（RTS）, the game agent's ability to compete against each other has undergone a subversive leap. Due to the high similarity between RTS games and air defense operations, how to successfully apply the game agent technology to air defense weapons is worthy of in-depth study and consideration. In the future battlefield environment of large-scale, highly dynamic and strong confrontation, in order to equip air defense weapons the ability to make intelligent decisions like human beings, an analogy thinking between air defense operations and RTS games is proposed, the successful experience and key technologies of RTS game agents are drawn. The intelligent decision-making architecture training of air defense weapon , the intelligent decision-making architecture of single-equipment and intelligent decision-making architecture of multi-equipment and algorithm are studied in detail, aiming to introduce the artificial intelligence technology into the air defense weapon equipment to achieve the purpose of improuing the combat effectiveness of weapon systems.