Consensus Of Multi-agent Systems Based On Triggering And Quantization Mechanisms

With the recent development of automation technology,multi-agent systems have been applied in more fields than ever,and also faced with various harsh operating environments and more severe conditions,primarily channel bandwidth limitations and transmission delays,absence of absolute information due to inability to obtain GPS signals,and the on-board energy limit of a single agent.These factors greatly limit the performance and even effectiveness of multi-agent systems.In order to make the multi-agent system operate normally under the influence of these unfavorable factors,this dissertation designs the corresponding control strategies and obtains ideal result.The main contents and innovations are as follows:(1)When GPS signal is not available,the multi-agent system cannot obtain absolute state information,the only available information is the measured relative state(edge state)information.Under such condition,conventional event-triggered control strategy which relies on absolute information is no longer suitable.To address this problem,in this dissertation,an edge triggered rule based on relative state and time function is designed for single-integrator and double-integrator multi-agent systems.It is proved that this triggering rule can also avoid Zeno phenomenon.Case with different time-varying transmission delays is further considered,and a novel distributed event-triggered sampling data transmission strategy and protocol was designed to solve the consensus problem.The LMI-based sufficient conditions for consensus is derived from the designed Lyapunov-Krasovskii functional.(2)Finite-time and fixed-time controller can achieve precise convergence within a limited time(or fixed time)and thus have been applied to multi-agent systems in recent years.In two recent papers,event-triggered control strategies were used to prevent continuous updating of controllers in the limited-time and fixed-time conformance control of multi-agent systems,and still achieve the precise consensus without Zeno behavior.Although Zeno behavior is ruled out in most cases,it cannot be avoided as the multi-agent system approaches final consensus.In this paper,the inevitability of Zeno behavior is proved by theoretical analysis and numerical simulation.Improved event-triggered control strategies are proposed to achieve practical consensus under finite-time and fixed-time dynamics,Zeno behavior is also excluded.(3)Quantization exists in all computer-based systems(such as multi-agent systems),this dictates that multi-agent systems can only achieve consensus of quantized values instead of precise consensus.Roger W.Brockett and Daniel Liberzon designed dynamic quantizers for linear systems that allows for precise consensus.In our work,the aforementioned dynamic quantizer is improved to reach faster convergence rate,then the improved dynamic quantizer is applied to first-order and second-order multi-agent system,and achieved desired results.This paper also makes several improveme'nts to the LMI-based dynamic quantization strategy proposed by Kun Liu and Emilia Fridman,hence significantly improved the convergence rate and reduced the estimation of convergence speed.Based on this strategy,a dynamic quantization strategy for multi-agent systems with multiple time-varying delays and asynchronous sampling is proposed,and desirable results are achieved.(4)AUV(Autonomous Underwater Vehicle)multi-agent systems consume considerable amount of energy on their underwater communications,continuous-time communications would heavily stress the on-board batteries,hence sampled time communication is used instead.However,the stability of the existing AUV multi-agent system sampled-time control is proved to be flawed,i.e.an important result that apply only to the individual sampling interval is erroneously applied to the entire time domain,thus the proof is not valid.The existing proofs are analyzed and a new proof method based on small gain theorem is proposed,which guarantees the mathematical rigorousness of the proof while giving a new stability criterion.This method also paves the way for the future implementations of event-triggered strategies on AUV multi-agent system.