Search

This paper investigates the adaptive consensus with limited information for a class of uncertain nonlinear multi-agent systems (MASs). The limited information, meaning the incomplete information transmitted between agents, stems from nonzero-kernel weight matrices of communication networks, which deserves concerns owing to its practical interest such as in privacy preserving and information compression. Apart from the limited/incomplete information, the paper considers nonlinear MASs with large uncertainties (without known upper bounds). This is essentially different from the existing results of linear systems and poses challenges for the design and analysis of adaptive consensus. As for fully distributed protocol design, dynamic high gains are introduced to compensate for the system uncertainties and for the unavailable global graph information. In closed-loop consensus analysis, a new Lyapunov function is constructed by utilising weight and incidence matrices instead of Laplacian matrix. It turns out that the designed protocol can guarantee the boundedness of closed-loop MASs signals and the convergent consensus. Two numerical examples are given to demonstrate the effectiveness of the proposed control scheme. [ABSTRACT FROM AUTHOR]

Communication network could suffer uncertainties originating from link faults and network attacks. This situation, in multi-agent systems, can be featured by unknown weights of network topology. To impair the influence of the uncertain topology, typical compensation should be taken into account in distributed protocols to the workability of multi-agent systems. This paper, in the context of uncertain topology, addresses finite-time leader-following consensus for second-order uncertain nonlinear multi-agent systems. In addition to uncertainties in topology, the systems also allow unknown control coefficients, which together with the unknown weights renders the realisation of finite-time leader-following consensus nontrivial. Specifically, a fully distributed protocol based on distributed finite-time observer is designed via integrating the adaptive compensation scheme. Notably, in the designed protocol, dynamic high gains are introduced for the compensations of the network uncertainties and agents uncertainties. It turns out that the designed fully distributed protocol guarantees the global finite-time leader-following consensus. Simulation examples are provided to illustrate the validity of the proposed approach. [ABSTRACT FROM AUTHOR]

Event-triggered control, as networked systems are popular, is increasingly appreciated. Such control could arouse essential saving of communication/computation and would lay the groundwork for networked (control) systems. However, due to the discontinuity and hybrid nature of event-triggered control, the exclusion of Zeno phenomenon is the premise of its workability. Aiming to circumvent the Zeno behaviour more explicitly, we attach a dynamic dwell time to triggering mechanism and propose a new adaptive event-triggered control scheme for uncertain nonlinear systems. Specifically, a set of dynamic gains are first introduced to compensate for large uncertainties and nonlinearities, and the indispensability of a set of gains rather than a gain is elaborated on. Then, from a dynamic-gain-dependent monitoring rule, the positive dwell time is generated so that the Zeno phenomenon can be excluded automatically. In particular, the rule enables system behaviour to be always monitored by evaluating the sampling errors of gains. Based on this, we construct an adaptive controller with linear structure, which allows the execution error can be estimated explicitly, and in turn suppressed by the gains. It is shown that, with the proposed event-triggered control, all the closed-loop system signals are bounded and the system state ultimately converges to zero. Notably, the designed triggering mechanism could save more resources in theory since the event detection and gains monitoring are performed alternately. The effectiveness and superiority of the proposed scheme is illustrated by two simulation examples. [ABSTRACT FROM AUTHOR]

In this paper, we investigate global-exponential stabilisation of a class of nonlinear networked control systems. The network-induced delays are assumed to be random and significantly smaller than the sampling period. First, sufficient conditions are presented to ensure global-exponential stability for a class of hybrid systems with time delay. Then, the networked control systems are modelled as the hybrid systems with time delay. By the techniques of adding a power integrator and a recursive argument, a sampled-data state feedback control law is presented. Sufficient conditions are given to ensure global-exponential stability of the closed-loop system by constructing a Lyapunov–Krasovskii function. Finally, a numerical example is presented to show the validity of the new methods. [ABSTRACT FROM AUTHOR]