Scafuti, Francesco (2016) Evolving and adaptive strategies for consensus and synchronization of multi-agent systems. [Tesi di dottorato]


Download (3MB) | Preview
[error in script] [error in script]
Item Type: Tesi di dottorato
Lingua: English
Title: Evolving and adaptive strategies for consensus and synchronization of multi-agent systems
Date: 31 March 2016
Number of Pages: 110
Institution: Università degli Studi di Napoli Federico II
Department: Ingegneria Elettrica e delle Tecnologie dell'Informazione
Scuola di dottorato: Ingegneria dell'informazione
Dottorato: Ingegneria informatica ed automatica
Ciclo di dottorato: 28
Coordinatore del Corso di dottorato:
di Bernardo, MarioUNSPECIFIED
Date: 31 March 2016
Number of Pages: 110
Uncontrolled Keywords: evolving networks; synchronization; consensus
Settori scientifico-disciplinari del MIUR: Area 09 - Ingegneria industriale e dell'informazione > ING-INF/04 - Automatica
Date Deposited: 03 May 2016 13:04
Last Modified: 31 Oct 2016 11:16


We investigate evolving and adaptive strategies, in network of dynamical agents, for solving general types of consensus and synchronization. First, we analyse the problem of max/min consensus in directed networks of integrators. Extending edge snapping method with a three-well potential, we are able to show the effectiveness of our strategy to achieve general types of consensus, different from the average. Theoretical results are validated via a number of numerical examples. Then we move to synchronization of coupled non identical oscillators. We design an evolutionary strategy for network synchronization. Our results suggest that heterogeneity is the driving force determining the evolution of state-dependent functional networks. Minimal emergent networks show enhanced synchronization properties and high levels of degree-frequency assortativity. We analyse networks of N = 100 and N = 1000 Kuramoto oscillators showing that hubs in the network tend to emerge as nodes' heterogeneity is increased. Finally, we study synchronization of multi-agent systems from a contraction theory viewpoint. Contraction theory is a useful tool to study convergence of dynamical systems and networks, recently proposed in the literature. In detail, we recall three strategies: virtual systems method, convergence to a flow-invariant subspace and hierarchical approach. While the former is simple to apply, the latter is suited for larger networks.

Actions (login required)

View Item View Item