Fuzzy logic controller (FLC) has become populer after receiving many succesful reviews in a wide of industrial applications. These range from washing machines to trains, from cement kilns to nuclear reactors.
Although FLC has been used in many succesful application, some classical control egineers are very sceptical of the technologi. Some of their reservation about FLC involves the following points.
Is FLC as simple and easy to design as is reported?
How does the performance of an FLC compare with conventional control algorithms?
Is FLC as robust as conventional control?
The aims of this thesis is to use genetic algorithms to answer the above questions. This is to be achieved by using genetic algorithms to tune both conventional controllers and FLCs. In carrying out this research, the issues of optimality of performance convergence and robustness are addresedd.
Moreover, in the light of long convergence times when using the genetic algorithms to tune FLCs, a simplified way to design FLCs based on the structure of the conventional PI (proportional plus integral) controller is synthesized. The resulting PI-Based FLC is easy to tune and provides better performance than conventional PI control.