The cognition and reasoning of the battlefield environment is an important ability of the unmanned systems’ autonomous capabilities. In this paper, a semantic-based knowledge representation framework for mission cognition of autonomous unmanned systems is proposed. Firstly, the attributes of entities in the battlefield are semantically represented. Then, the Recognition Graph for Situation Awareness (RGSA) is proposed and used to connect the entities through semantic information, which provides a foundation for cognition and reasoning. Furthermore, the SWRL(Semantic Web Rule Language) rules can be utilized to generate the local planning situation windows or achieve event perception, which imitates the human attention mechanism and event perception. Finally, a simulation test system is constructed, and the test results show that the semantic-based knowledge representation framework is useful for the situation perception and autonomy of unmanned combat systems.