Cluster Evaluation Method Based on Multi-agent Deduction and Complex Network Technology

UAV cluster technology is widely used in logistics, military, agriculture, and other fields. It is an important direction for future development. Because of the characteristics of high cluster complexity, strong information interaction capabilities, and variability in application fields, the analytical evaluation of cluster systems are increasingly popular. The focus of attention. Due to the UAV cluster's multi-level and multi-element complex interaction, the abstract evaluation of the cluster system using the complex network topology is critical. But the actual task process of the cluster system is variability, emergence, and autonomy. Such as dynamic performance, it isn't easy to comprehensively analyze the entire task process only by relying on the static complex network topology, and the application parameters evaluated are relatively theoretical. With the continuous development of simulation technology, the multi-agent simulation modeling method is very suitable for simulating complex group behaviors with unknown states and emergence due to the interaction characteristics of multi-state independent agents. Applying the multi-agent simulation modeling method to the field of UAV clusters can solve the problem that the existing complex network methods are challenging to simulate the actual task process of the cluster. Therefore, this paper proposes a method of using multi-agent simulation to provide dynamic parameters. This method combines the evaluatssion and analysis method of complex network topology to evaluate the system-level network indicators of the drone cluster task process.