| Summary: | Networked Control Systems (NCS) and Wireless Networked Control Systems (WNCS) are control systems where controllers, sensors and final elements of control are connected to a mutual communication network. The inclusion of the network introduces delays and dropouts, which greatly influence the stability and robustness of the controller. While there is wealth in theoretical contributions to NCS, it is still imperative to study more applications and investigate the effects of networks in a real-time operation. There are also open problems that require further study of the impact of disturbances, constraints and strong interactions in complex NCS. This thesis is concerned with the design of control strategies for WNCS mainly focused on Model-Based Predictive Control (MBPC), Proportional Integral Derivative (PID) and decentralised schemes with the aim of creating control laws suitable for compensating time-varying delays and dropouts. These strategies rely on optimisation problems which incorporate robustness and performance restrictions to compute the optimum controller. The performance and robustness of the controllers are evaluated through extensive experiments in a network simulator. A new adaptive Internal Model Control (IMC) controller has been developed to adapt to the network requirements and compute the IMC model parameters online. A new robust PID for NCS under random delays has been created by solving a new constrained optimisation problem that included constraints of maximum sensitivity to guarantee robustness. A novel optimal immune PID is developed to improve the performance of NCS under time-varying delays and dropouts. Simulation results show that the controller offers greater flexibility and improves the performance and robustness with respect to the other methods studied. Four more controllers have been tested and extensive tests have indicated stability for a limited percentage of process model variations and dropouts. Predictive PID controllers, with similar properties to MBPC, are developed to compensate dropouts in WNCS. A quadratic programming problem optimises a new MBPC cost function to find the optimal PID gains. The approach successfully maximises the performance by changing the controller gains at every sampling time and allowing maximum variations of system parameters and dropouts. Also, a new constrained predictive PID controller is presented to deal with input saturation. Simulation results show the superiority of the design in comparison with the control schemes studied earlier. Furthermore, a decentralised wireless networked model predictive control design for complex industrial systems has been developed. Also, the method has been applied to wind farm control. The proposed decentralised control offers an effective and innovative solution to improve the performance of large industrial applications.
|
|---|
