Reliable Memory Sampled-Data Consensus of Multi-Agent Systems With Nonlinear Actuator Faults

Abstract

This brief proposes a memory-based sampled-data consensus framework for general linear multi-agent systems (MAS) in the presence of a class of nonlinear actuator faults (NAF). To reduce state exchanges and preserve energy resources, communication between the neighboring agents are based on only samples of the states with variable sampling intervals. As two common constraints in the actuators, the bounded nonlinear partial loss of effectiveness and bias faults are both taken into account in the problem formulation. Sufficient conditions to guarantee consensus under the given circumstances are derived as linear matrix inequality (LMI) conditions. Different from existing Lyapunov-Krasovskii-based methods, the proposed design framework in this brief is based on a looped functional approach which reduces the conservation in designing the required consensus control gains. This less conservative approach allows a larger sampling interval as well as more severe actuator faults which together enhance the practicability of the proposed approach. Simulation results based on a tunnel diode circuit and a non-holonomic mobile robot MASs quantify the effectiveness of the proposed approach and the improved sampling intervals.