Robust control design using modern constrained optimization techniques /
Robust control design is commonly a difficult task that requires complicated mathematical formulation and heuristic parameters tuning. In addition, if often results in a high order controller. Motivated by the need to reduce complexity, a robust state feedback control design using modern constrained...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
Kuala Lumpur :
Kulliyyah of Engineering, International Islamic University Malaysia,
2012
|
| Subjects: | |
| Online Access: | http://studentrepo.iium.edu.my/handle/123456789/5191 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Robust control design is commonly a difficult task that requires complicated mathematical formulation and heuristic parameters tuning. In addition, if often results in a high order controller. Motivated by the need to reduce complexity, a robust state feedback control design using modern constrained optimization algorithms is proposed in this thesis. Combining the advantages of robust control theory and computational intelligence makes the task more straightforward and automatic. Basically, a robust control design requires a set of goals to be achieved such as good transient response, zero steady state error for a constant input and most importantly, robustness to parameter uncertainty. A single-objective constrained optimization technique is used in the proposed method to handle these requirements. Searching for a set of robust controller gains that maximizes the stability radius of the closed-loop system is the objective. The constraint of the optimization is the region for the closedloop poles that represents the desired time-domain control performance. In the beginning, the study is focused to find the suitable modern optimization tool(s) among the commonly used optimization tools such as Genetic Algorithm, Particle Swarm Optimization and Differential Evolution. The study further investigates the optimization features, such as constraint handling, stopping criterion and choice of optimization parameters. The result shows that Differential Evolution (with mutation factor=0.5 and crossover constant=0.9) outperforms Clerc's Particle Swarm Optimization and Genetic Algorithm in constrained optimization problems. At the end of the study, the proposed robust control design using Particle Swarm Optimization and Differential Evolution are applied to pendulum-like systems, such as gantry crane, flexible joint and inverted pendulum. A set of laboratory experiments are carried out to evaluate the performance of the designed controller. LQR-based controller and loop-shaping controller are also designed for comparison with the proposed controller. The advantage of the proposed controller design is the automated tuning process for the controller parameters as compared to the benchmark controllers. Another contribution of the thesis is the dynamics modeling of the pendulum-like systems where a generic model structure for the pendulum-like systems is developed. The generic model structure is obtained by linearization around equilibrium and simplification where the dynamics effect of vibration to actuator dynamics is neglected. As a result, the parameters of the pendulum-like systems model can be easily identified by decoupling of vibration model and actuator model in the experiment. |
|---|---|
| Item Description: | Abstracts in English and Arabic. "A thesis submitted in fulfilment of the requirement for the degree of Doctor of Philosophy in Engineering."--On t.p. |
| Physical Description: | xxiii, 178 leaves : ill. ; 30cm. |
| Bibliography: | Includes bibliographical references (leaves 160-168). |