Use this URL to cite or link to this record in EThOS:
Title: Analysis and implementation of time-delay systems and networked control systems
Author: Wang, Bo
ISNI:       0000 0001 3558 4868
Awarding Body: University of Glamorgan
Current Institution: University of South Wales
Date of Award: 2008
Availability of Full Text:
Access from EThOS:
Access from Institution:
Systems with delays frequently appear in engineering. The presence of delays makes system analysis and control design much more complicated. Networked control systems where the delays are often random are typical cases of such systems. For one particular category of time-delays systems, integral processes with dead time (IPDTs), the control limits that a PI controller can achieve are discussed in this thesis. These limits include the region of the control parameters to guarantee the system stability, the control parameters to achieve the given gain and/or phase margins (GPMs), the constraint on achievable gain and phase margins, the performance of set point tracking and disturbance rejection. Three types of PI controllers, namely typical PI controller, single tuning-parameter PI controller and PI controller under two-degree-of-freedom (2-DOF) structure, are studied. In control schemes of the modified Smith predictor (MSP) where the controller usually includes a distributed delay, the system implementation is not trivial because of the inherent hidden unstable poles. This thesis provides an estimation of the minimal number of implementation steps for the distributed delay in linear control laws. This is obtained by solving an inequality with respect to the number of implementation steps. A coarse estimation is given as the initial value to solve the inequality using bisection algorithms. A minimization process as well as some other techniques are also introduced to further improve the estimation. In networked control systems, the network-transmission delay and data dropout are combinedly represented by a network-induced delay. By designing a data pre­ processing mechanism, the network-induced delay can be assigned. Such delay as­ signment is applied to networked predictive control schemes, which alleviates systemstability limits on the network-induced delay. Two stability criteria are given for the closed-loop system with random network-induced delay, and a resulting implementation algorithm is also provided. The control and implementation of a magnetic levitation system over the network is studied in this thesis. Firstly, a test-rig which is suitable to implement control over a network is set up. Feedback linearization and direct local linearization methods for the nonlinear MagLev system are presented. In order to improve the control performance, a networked predictive method is employed, where the system model is identified in real-time. Local control and networked control are implemented on this test-rig, including networked predictive control. Model predictive control demonstrates a clear performance advantage over the networked control strategies which does not incorporate compensation for the network-induced delay. In order to quickly implement networked control systems (NCSs) by simulation or practical application, a MATLAB/Simulink based NCS toolbox is developed. This toolbox incorporates basic parts of a general NCS, that is, network simula­tion, network interface, plant interface and typical control schemes. With the NCS toolbox, users can focus on the study of new control schemes.
Supervisor: Liu, Guoping Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: Computer networks ; Automatic control