Unmanned swarm combat is a typical combat paradigm in future intelligent warfare. We focus on the higher-order topological structure of unmanned swarms and study how the information networks formed by information and control flows within different topological structures change in collaborative capabilities when facing complex collaborative tasks. Based on the association rules of the swarm intelligence of unmanned swarms, we design the unmanned swarm information network as a modular community network, constructing four types of intra-module network models including nearest-neighbor coupled networks, random networks, small-world networks, and scale-free networks, and two types of inter-module relationships, random connections, and preferential connections. By using hypergraphs to construct collaborative hyperedges composed of task collaboration, and combining topological indicators of higher-order networks such as hyper-degree, hyper-degree distribution, and synchronization index, we comprehensively evaluate the quality and efficiency of swarm collaboration. Simulation experiments are conducted, and the results show that the preferential connection small-world network model can better balance collaboration quality and efficiency under the given tasks, enhancing collaborative capabilities.