Abstract: For the shortage of plan recognition method presented by Kautz(1987) in controlling mechanism, a plan recognition method based on planning knowledge graph is discussed in detail. Using first-order predicate logic as formal language, an instance of tactical intelligent planning is analyzed and a situation assessment model combined tactical intelligent planning system and a tactical plan recognition system are proposed. The plan library, which is dynamic set up, could break through the limit of closed world hypothesis in Kautzs plan recognition method. The model is capable of supplying reasonable hypothesis for enemys action goal and plan and projecting enemys future action, which is helpful to command decision.