Abstract
In this article, a distributed model-free consensus control is proposed for a network of nonlinear agents with unknown nonlinear dynamics, unknown process disturbances, and white noise measurement disturbances. Here, the purpose of the control protocol is to first synchronize the states of all follower agents in the network to a leader and then track a reference trajectory in the state space. The leader has at least one information connection with one of the follower agents in the network. The design procedure includes adaptive laws for estimating the unknown linear and nonlinear terms of each agent’s dynamics. The salient feature of the proposed control scheme is that each agent’s estimation is a model-free adaptive law, that is, the need for regressor or linear-in-parameter basis is alleviated. In addition, without requiring direct connection to the leader, the leader’s control input can still be reconstructed by virtue of a robust observer which can be defined in a distributed manner in the network. The entire design procedure is analyzed successfully for the stability using Lyapunov stability theorem. In addition, it is shown that the proposed distributed controller includes an optimal term. Besides, a modified Kalman filter is added to eliminate the measurement noise. Finally, the simulation results on three networks of unknown nonlinear systems are presented. Moreover, a comparative study is presented to evaluate the proposed algorithm against a model-based cooperative control algorithm.
Highlights
Great attention has been paid to the problem of controlling multiagent systems ranging from consensus to formation control.[1,2,3,4] The solutions applied to oscillator synchronization, mobile robot and aircraft formation, mobile sensor area coverage, vehicle routing in traffic, containment control of moving bodies, and so on.[5]
All of these problems can be considered as a consensus problem, in which all agents’ states should be synchronized inside a network.[6]
Several model-free controllers (MFCs) have been proposed for a School of Electrical and Electronic Engineering, Universiti Sains Malaysia, Nibong Tebal, Pulau Pinang, Malaysia
Summary
Great attention has been paid to the problem of controlling multiagent systems ranging from consensus to formation control.[1,2,3,4] The solutions applied to oscillator synchronization, mobile robot and aircraft formation, mobile sensor area coverage, vehicle routing in traffic, containment control of moving bodies, and so on.[5]. Distributed adaptive leader-following control for unknown dynamic systems with guaranteed finite-time convergence is proposed by Mahyuddin et al.[18,19] The algorithms are model-based cooperative controllers which require sufficiently rich input signals to guarantee persistently excitation condition for the regressors. A distributed consensus control problem is solved for a network of agents with general unknown nonlinear multi-input and multi-output (MIMO) dynamic system using a model-free control algorithm. The main contribution of this article is the design and development of an MFC algorithm for consensus problem involving a network of nonlinear multiagent systems without requiring the use of ANNs to estimate the unknown system dynamics and disturbances. That section includes three different subsections dedicated to distributed estimation for unknown system matrix, adaptive MFC cooperative protocol, and cooperative robust observer for the leader’s control inputs.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
