The article promoted a kind of Unmanned Aerial Vehicle (UAV) system design architecture with AI training environment using Model Based System Engineering (MBSE). The system provides framework base on system of system design and AI algorithm verification with training environment, which covers the whole design life process of UAV system. First of all, the architecture of the UAV system was designed and analyzed base on Universal Architecture Framework (UAF), by following the digital simulation in time-space domain. The result of system simulated and evaluated help to improve the system design. Furthermore, AI training environment was integrated with the time-space domain simulation so that UAV system can be validated in adversarial scenarios, which obviously more close to real world. Besides, the avionics hardware also are integrated to as Hardware in Loop (HIL) verification for whole system, especially the components of UAV under the test. Finally, three dimension visualization simulation is also involved into the system integration by building up the Live Virtual and Constructive (LVC) simulation environments. Base on the system mentioned by the article, UAV system can be designed and verified according to the requirement strictly. What’s more, the AI self-learning environment can be integrated since the system can provide simulation data for training. The whole system design stages are covered by MBSE from requirements analysis to system logical and function verification, which also can be used in the digital twins in the future.