Research On The Consensus Problems Of Networked Systems

Networked systems are composed of agents with the ability of sensing,computing,communicating and decision making.With the distributed interaction and cooperation among the agents,networked systems can perform some sophisticated tasks which are difficult or even can not be solved by a single agent.Networked systems with all these characteristics have comprehensive engineering applications and become the research focus in recent years.This thesis also focuses on the cooperation control of networked systems.We investigate the semi-global consensus of multi-agent systems subject to input saturation and edge consensus of complex networks.The main work can be summarized as follows:1.Investigation on semi-global consensus tracking of multi-agent systems subject to input saturation and periodically intermittent feedback controllers.In periodically intermittent feedback control,each agent only works for a fraction of time during a given period and rests in the remaining of time.For general linear systems with each agent being asymptotically null controllable with bounded controls,distributed periodically intermittent feedback controllers are first constructed by utilizing the algebraic Riccati equation(ARE)based low-gain feedback technique.Then,sufficient conditions are provided to guarantee the semiglobal consensus tracking.2.Investigation on event-triggered semi-global consensus of multi-agent systems subject to input saturation.In event-triggered strategy,the sensor and actuator of each agent will not sample or update information until the designed events have been triggered.For general linear systems with each agent being asymptotically null controllable with bounded controls,distributed consensus protocol is constructed on the basis of ARE-based low-gain feedback technique and sufficient conditions are derived for achieving semi-global consensus.Furthermore,the case that there exists time delay between the information sampling and information updating is also been analyzed.For the case without or with updating delay,it is proved that the Zeno behavior can be excluded.3.Investigations on semi-global consensus of networked systems with input saturation together with dead zone and input additive disturbance.Distributed consensus protocol referring to the relative state information is first constructed.By using the parameterized low-and-high gain feedback technique,sufficient conditions guaranteeing the semi-global robust consensus and the semi-global robust swarm of undirected connected multi-agent systems with each agent being asymptotically null controllable with bounded controls are provided,respectively.Then,distributed observer-based consensus protocol is designed and sufficient conditions are provided,respectively,to guarantee the semi-global robust consensus and the semi-global robust swarm for undirected connected multiagent systems with each agent being asymptotically null controllable with bounded controls,detectable,left-invertible and minimum-phase,by combining the parameterized low-and-high gain feedback technique and the special coordinate basis decomposition theory as well as the high-gain observers design approach.4.Investigation on edge consensus of complex networks with linear edge dynamics.Distributed discrete-time and continuous-time edge consensus protocols are proposed for general linear systems and conditions are derived respectively for these two kinds of protocols to guarantee the achievement of edge consensus.Furthermore,the influence of network structure on the speed of reaching edge consensus together with the heterogenicity and the degree assortativity property of the corresponding line graph is analyzed,by taking a scale-free network with a tunable exponent ? between 2 and ? as examples.5.Investigation on nonnegative edge consensus of networked systems.Convergence and nonnegativity of edge states are both taken into consideration in this theme.On the basis of the property of Metzler matrix and Lyapunov stability theory,distributed sufficient conditions are derived for the nonnegative edge consensus of networked systems.The low-gain feedback technique used in systems with input saturation is introduced to simplify the design of the feedback gain matrix for achieving the nonnegative edge consensus.