摘要：In this paper, we deal with a hierarchical control system composed of a large-scale heterogeneous consensus multi-agent system and an external controller with weak feedback. We consider two issues; (1) find the controller design strategy in the upper layer without the information on the transfer functions of all the agents, (2) find the optimal choice of controlled/observed agents for the convergence rate of the whole system with low calculation costs compared to the brute force algorithm. We propose a passivity-based controller design for the upper layer and show that the optimal choice of the controlled/observed points is given by an eigenvector of a matrix which represents the network structure.
