Adaptive Tracking Control for Uncertain Nonlinear Multi-Agent Systems With Partially Sensor Attack

Abstract

Although rich collection of research results on sensor attack (SA) exist, no attack detection mechanism has ever been designed based on output error information to detect whether an attack has occurred. The primary objective of this article is to build a backstepping adaptive tracking control protocol for heterogeneous nonlinear uncertain multi-agent systems (HNUMASs) with partially SA. A dynamic SA detection mechanism by using output error information of agent only is developed to identify the SA. After locating the attack, to circumvent the effects of unknown time-varying output gain caused by SAs, we introduce the Nussbaum function in backstepping design to compensate the unknown time-varying output gain. The result shows that the developed SA detection mechanism is able to detect the occurrence of attack in a timely manner, while the derived adaptive backstepping tracking controller can effectively handle the adverse effects of SA and ensure that all signals are bounded in the closed-loop system. At last, the effectiveness and benefits of the presented approach are verified by simulation example. Note to Practitioners—This paper aims to achieve the adaptive tracking control for HNUMASs under SA, which can be widely used in practice, such as power systems, vehicular platoon systems, etc. The control protocol consists of a attack detection mechanism and Nussbaum function that compensates for the unfavorable effects caused by SAs. Moreover, the system may also be influenced by uncertainties from its neighboring agents in practical applications. Therefore, an additional estimator is designed in each subsystem to handle the uncertainties involved in its neighbor dynamics. This design avoids the exchange of information related to local neighborhood consensus errors among connected subsystems. A feasible strategy is provided for industrial applications.