1 January 2018 in Complexity
It is well known that all agents in a multiagent system can asymptotically converge to a common value based on consensus protocols. Besides, the associated convergence rate depends on the magnitude of the smallest nonzero eigenvalue of Laplacian matrix L. In this paper, we introduce a superposition system to superpose to the original system and study how to change the convergence rate without destroying the connectivity of undirected communication graphs. And we find the result if the eigenvector x of eigenvalue λ has two identical entries xi = xj, then the weight and existence of the edge eij do not affect the magnitude of λ, which is the argument of this paper. By taking advantage of the inequality of eigenvalues, conditions are derived to achieve the largest convergence rate with the largest delay margin, and, at the same time, the corresponding topology structure is characterized in detail. In addition, a method of constructing invalid algebraic connectivity weights is proposed to keep the convergence rate unchanged. Finally, simulations are given to demonstrate the effectiveness of the results.
4 August 2020 in IEEE Transactions on Cybernetics
In this article, by analyzing the eigenvalues and eigenvectors of Laplacian L, we investigate the controllability of multiagent systems under equitable partitions. Two classes of nontrivial cells are defined according to the different numbers of links between them, which are completely connected nontrivial cells (CCNCs) and incompletely connected nontrivial cells. For the system with CCNCs, a necessary condition for controllability is found to be choosing leaders from each nontrivial cell, the number of which should be one less than the cardinality of the cell. It is shown that the controllability is affected by three factors: 1) the number of the links between nontrivial cells; 2) the rank of the connection matrix; and 3) the odevity of the capacity of the nontrivial cells. In the case of nontrivial cells under the equitable partition, there are automorphisms of interconnection graph G, which induce the eigenvectors of L with zero entries. For the system with automorphisms, by taking advantage of the property of eigenvectors associated with L, we propose several graphical necessary conditions for controllability. In addition, by the PBH rank criterion, the controllable subspaces of the system with different classes of nontrivial cells are compared. Finally, a necessary and sufficient condition for controllability under minimum inputs is given.
15 April 2020 in IEEE Access
This paper studies the controllability of leader-follower multi-agent systems under fixed topology. The relationship between controllability and information communication topologies is characterized by taking advantage of equitable partitions. Firstly, for one non-trivial cell taking leaders role, a graph theory method is proposed to determine the multi-agent controllability. Secondly, the controllability of undirected graphs consisting of four nodes is analyzed in detail, and the limitation of the equitable partition in controllability analysis is pointed out for the first time. Thirdly, the results of four-node graphs are extended to general topologies, by which, we reveal the relationship between the ranks of controllability matrix of the quotient graph and original graph. Finally, the effect of non-trivial cells on the controllable subspace is analyzed from the viewpoint of controllability matrix and eigenvector.
17 July 2024 in International Journal of Control
7 March 2025 in IEEE/CAA Journal of Automatica Sinica
The multi-agent controllability is intrinsically affected by the network topology and the selection of leaders. A focus of exploring this problem is to uncover the relationship between the eigenspace of Laplacian matrix and network topology. For strongly connected directed circle graphs, we elaborate how the zero entries in the left eigenvectors of Laplacian matrix L arise. The topologies arising from left eigenvectors with zero entries are filtered to construct essentially controllable directed circle graphs regardless of the choice of leaders. We propose two methods for constructing a substantial quantity of essentially controllable graphs, with a focus on utilizing essentially controllable circle graphs as the foundation. For a special directed graph-OT tree, the controllability is shown to be related with its substructure-paths. This promotes the establishment of a sufficient and necessary condition for controllability. Finally, a method is presented to check the controllable subspace by identifying the left eigenvectors and generalized left eigenvectors.
24 July 2024 in IEEE Transactions on Control of Network Systems
In this paper, we discuss the controllability of reachable nodes in directed graphs over consensus protocol. A node is called a reachable node if there is a path from an input to this node, otherwise unreachable node. The rows in the controllability matrix associated with unreachable nodes are shown to be zero rows, and a prerequisite for controllability is that all nodes are reachable. A method for constructing controllable directed graphs is provided later in the paper. If the in-degrees of all nodes are distinct, and there are neither sibling nodes nor unreachable nodes, the system is controllable. A subsystem composed of reachable nodes is called a reachable subsystem. All reachable nodes are proved to be controllable if and only if the reachable subsystem is controllable. For a reachable subsystem with a tree graph, a necessary and sufficient condition for controllability is given. Besides, if the reachable subsystem with a tree graph is uncontrollable, a graphical method is given to identify the dimension of controllable subspace.