Analysis and synthesis of networked control systems (Q533283)

The authors discuss four problems related to Network Control Systems (NCSs) i.e., quantization, data fusion, predictive control and fault detection. NCSs are understood as distributed control systems closed via communication links with delay and bandwidth constrained. A quantizer is used to map real-valued signals into piecewise constant ones with a finite set of values which are sent to the network. The authors study the problem of mean square stability of linear discrete-time quantized systems taking into account packet dropout. In the first stage sufficient conditions of the stability of the closed-loop system which contains the Kalman filter as a state estimator, a simple quantizer which transforms the estimated states into a sequence of integers using \(n\)-dimensional super square boxes and a controller which acts linearly on the decoded values. The stability analysis is based on Lyapunov type reasoning. To cope with the problem of packet dropout, the authors propose the ``zoom'' strategy based on the assumption that changing the sensitivity and saturation parameters of the quantizer is possible. It enables to formulate sufficient conditions for mean square stability of the closed-loop system which in turn may be used to find the largest packet dropout which does not violate the stability property. The results are then extended to systems with quantized inputs computed from quantized measurements. In this case, logarithmic quantizers are assumed and stability analysis is based on the use of a nonnegative bounded multiplier. The problem of multiple sensors that sample asynchronously is discussed in the book. The authors propose the state fusion estimation algorithm based on the modified Kalman filter and test its behavior both analytically and using simulation experiments. Moreover, a new networked predictive controller is proposed which consists of the control prediction generator and the network data loss compensator. The efficiency of the proposed solutions is demonstrated on a number of examples solved numerically. Yet another problem considered in the book is related to fault detection in NCSs. The missing data of actuators are treated as noise and a method for the fault detection filters design is proposed. The authors propose also a compensation scheme for network predictive control systems with random time delay and clock asynchronism. All considerations in the book are illustrated by numerical examples. Although many of them have rather academic flavor but they demonstrate the efficiency of the proposed techniques and could be viewed as an introduction to realistic applications of the NCSs.