Moving Horizon Control And Estimation Of Networked Systems

In the last few decades, with the rapid developments of communication, computer and control technology, the traditional point-to-point control architecture cannot meet the required performance of control systems and changes to the networked control counterpart. Networked control systems (NCSs) have some nice advantages such as less wiring, low cost, easy maintenance and expanding. Due to these distinctive benefits, NCSs have found applications in a variety of areas. Therefore, NCSs have now been one of the hot research topics in the control society.However, although NCSs have brought so many advantages, they also bring lots of new problems and difficulties, such as network-induced delays, packet disordering, packet dropouts, quantization, et al. Though there have been fruitful research results on those issues, lots of problems remain to be unsolved. Especially, how to establish a model to de-scribe the phenomenon of packet disordering for NCSs with random or time-vary delay and relations between parameters characterizing the delays and performances? How to solve the optimal state estimation problem with system constraints and establish relations among the quantization density, the packet dropout probability and estimation performances? Based on the moving optimization strategy, the state estimation and controller design problem are respectively investigated for NCSs in this thesis. The main work is summarized as follows:1. A robust moving horizon H∞control method is proposed for NCSs with long time-delays, input constraints and disturbances. First, the delay is separated into a nominal part and a time-varying uncertain part, and then the NCSs are modeled as a class of uncertain systems with norm-bounded uncertainties. Based on the Lyapunov stability theory and moving optimization strategy, a sufficient condition is obtained such that the closed-loop system is asymptotically stable, and predictive controller is designed with a prescribed H∞performance level. The effectiveness of the proposed method is illustrated by a numerical simulation.2. A moving horizon H∞control method is proposed for NCSs with random delays and packet disordering. By introducing a logic data packet processor to reorder the packet from the network to actuator, the newest data signal sequence and a new transition proba-bility matrix are obtained, and then the NCSs are modeled as a class of Markovian jump systems. Based on the moving optimization strategy, state feedback controllers are de-signed by the proposed method. The impact of packet disordering on the performance of NCSs is effectively eliminated. The effectiveness of the proposed methods are illustrated by an angular positioning system.3. A switched moving horizon control method and a moving horizon H∞, control method are proposed for NCSs with time-varying delays and packet disordering, respec-tively. In the switched moving horizon control method, the NCSs are modeled as a class of switched systems by packet disordering. In the moving horizon H∞control method, the actuator reads the buffer periodically at a smaller period than the sensor, and the resulting system is a multi-rate system, and then converted into a parameter uncertain system with multi-step delay. Based on the proposed two methods, state feedback controllers are also designed, and relation between the delay variation rate and system performance is explicitly established for the NCSs.4. A moving horizon estimation method is proposed to deal with inequality constraints for networked systems with multiple packet dropouts. The packet dropout process is de-scribed a Bernoulli random process. By using hold-input strategy, the networked system is modeled as a stochastic parameter system model. Based on moving optimization strat-egy, the unconstrained optimization estimator is obtained with analytical solution. The design procedure for constrained estimator is presented for networked system with inequal-ity constraint of noises. Moreover, the stability properties of the estimator are studied, and a sufficient condition is obtained to ensure the estimation error to be norm-bounded. The effectiveness of the proposed method is illustrated by a simulation of a continuous stirred tank reactor.5. The moving horizon SINR estimation problem is investigated for wireless net-worked systems with random packet dropouts. Based on flow-rate and power control algo-rithm, a state-space model is obtained in dB scale for wireless networked system, and the bound constraints of SINR is derived and converted to state constraints. By using moving optimization strategy, the unconstrained SINR estimator is obtained with analytical solu-tion. Considering the coupling of state variable and process noise, a one-step MHE algo-rithm is presented to solve the constrained SINR optimization problem. The effectiveness of the proposed method is illustrated by a numerical simulation.6. The moving horizon estimation problem is investigated for networked systems with quantized measurements and packet dropouts. By using logarithmic quantizer and zero-input strategy for packet dropouts, the networked system is modeled as a stochastic param-eter uncertain system model. Based on moving optimization strategy, the optimal estima-tor and approximate estimator are obtained by solving a regularized least-squares problem with uncertain parameters. The stability properties of the optimal estimator are also stud-ied. The obtained stability condition implicitly establishes a relation between the upper bound of the estimation error and two parameters, namely, the quantization density and the packet dropout probability. Moreover, the maximum quantization density and the maxi-mum packet dropout probability are given to ensure the convergence of the upper bound of the estimation error sequence. The effectiveness of the proposed method is illustrated by a numerical simulation.