Model-Based Networked Control Systems

Networked Control Systems (NCS) are feedback control systems that use a data network as medium for the feedback path. NCS present a number of advantages over traditional wired-feedback control systems that include low-cost, easy maintenance, and flexibility. The targeted applications are industrial plants, geographically distributed systems, unmanned aerial vehicles (UAVs), etc. Thus, the study of NCS is of great interest but is technically complex because of the interaction between the continuous plant dynamics and the discrete and sometimes random nature of the network. This dissertation presents a class of networked control systems called Model-Based Networked Control Systems (MB-NCS). This control architecture uses an explicit model of the plant in order to reduce the network traffic while attempting to prevent excessive performance degradation. The model is used to simulate plant conditions between transmission instants when no information about the plant is available. It is assumed throughout the work that the plant model dynamics are a non-exact approximation of the real plant dynamics. Several control problems are formulated starting with the stability analysis of linear continuous and discrete systems. Necessary and sufficient conditions for stability are derived for different network and control scenarios such as state feedback, output feedback, and presence of networked-induced delays. The results obtained relate the stability of each of the MB-NCS considered to the eigenvalues of a test matrix. The test matrix contains information about the plant dynamics and the error between the plant and model dynamics. The structures of the test matrix for different MB-NCS are similar. This allows for a unified and systematic study of other more complex MB-NCS. The performance of linear continuous MB-NCS is also addressed using lifting techniques; two different types of H2-like performance indices are introduced and studied. This allows the designer to evaluate the performance of a specific controller under a fixed transmission rate. Next stability conditions are derived for MB-NCS with network time-varying transmission times. Popular models for network transmission times are used such as Markov-driven transmission times. Stability of quantized MB-NCS is also studied; sufficient conditions for uniform, logarithmic, dynamic quantizers are derived. Finally the stability properties of two classes of continuous non-linear plants are considered.