With the explosive growth of military data, the traditional static ontology based on expert knowledge has many disadvantages, such as strong subjective dependence on experts, poor portability, and poor learning ability on ontology. Which can no longer dynamically adapt to data changes and automatically update large-scale knowledge graphs. For military information applications, this paper proposes a dynamic ontology construction technique for the military field. The update of the ontology concept is completed through four parts: initial ontology design, data mapping, formal background merging, and ontology update. The technique can automatically update the ontology model with the change of information, facilitating the integration of multi-source heterogeneous data and the construction of knowledge graphs. Finally, taking the Russia-Ukraine conflict knowledge graph as a case study, the dynamic ontology technique can effectively discover new event types and related elements, realizing the ontology updated and fusion of multi-source heterogeneous data and the inference of military activity patterns based on the knowledge graph.