A distributed control framework and delay-dependent stability analysis for large-scale networked control systems with non-ideal communication network

Recently, the incorporation of telecommunication technology with control systems provides the usability of remote measurement signals for the designing of networked controllers for geographically distributed large-scale systems. However, communication network introduces new difficulties such as time delays and packet dropouts to the design of networked controllers. This paper presents a new Distributed Networked Control Scheme (DNCS) and its stability analysis for stabilizing of large-scale systems with interconnected subsystems featuring both random delays and random packet dropouts in their communication links. Firstly, a general model for large scale distributed networked system consisting of subsystems is used in which the state of each subsystem has its own time varying delay and there are also delays and packet dropouts in their interconnection communication links. To compensate the influence of subsystems on each other and enhance performance and stability margin of the closed-loop system, a suitable distributed control scheme is proposed. The stability criteria are provided based on Lyapunov-Krasovskii and Linear Matrix Inequality (LMI) techniques. For this, a new type of Lyapunov-Krasovskii functional and certain slack matrices are developed to conclude some LMI-based delay-dependent theorems for designing the control law, as well as the stabilizing of the DNCS. To evaluate the proposed method, three illustrative examples are provided. To indicate the efficiency of the suggested approach, a small-gain-based approach and an observer-based consensus method are applied for comparison. Simulation results show the effectiveness of the proposed approach to enhance the performance of large-scale networked systems among a non-ideal communication network.

LQR design of synchronizing controllers for multi-agent systems

The goal of synchronization of multi-agent systems is achieved, if for any initial condition the agents asymptotically follow a common trajectory. Besides asymptotic synchronization this paper considers the transient behavior of the synchronization errors between agents. A good transient behavior is expected for networked controllers that minimize an LQ-like performance index. For this type of objective functions the standard observability assumption of the linear quadratic regulator theory is not fulfilled and hence the optimization problem can not be solved directly. It is shown that the optimal networked controller requires an all-to-all coupling between the agents. To overcome the limitation of such a complete coupling, an approximation method is developed that ensures synchronization of the agents with a given structure of the communication network. For the case of a complete communication it is shown that the approximate networked controller is also optimal.