Consensus of Multi-Agent Systems via Structural Perturbation and Random Walk with Restart

Multi-agent collaborative control is widely applied in fields such as healthcare, traffic control, industrial automation, and military command. The consensus of multi-agent systems is a major research topic in cooperative control. Since low network connectivity can lead to inefficient collaboration of multi-agent systems, this paper proposes a new link prediction method aimed at optimizing the topology of multi-agent networks. The new method combines structural perturbation and random walk with restart, which can effectively utilize the edge and node information in the network to predict missing or possible future links. The new method can enhance the connectivity of the network and improve the collaborative efficiency of multi-agent systems. Comparison experiments with a single method show that the proposed new method significantly accelerates the convergence rate of the collaborative consistency in multi-agent systems.