Creative Design of Fuzzy Logic Controller
Lotfi Hamrouni, Adel M. Alimi · Kluwer Academic Publishers eBooks · 2006
The present paper proposes a novel idea used to conceive a proper fuzzy logic controller for an industrial application. Generally, the design of fuzzylogic controller is treated as numerical optimisation problem. Genetic algorithms and neural networks are the main techniques used to adjust the values of fuzzy logic controller parameters. The majority of approaches start with an empty fuzzy logic controller, having default parameters, and make improvement over the previous one. This operation is not really a design activity; it’sbetter called a learning process. In this work we regard the conception of a fuzzy logic controller as creative designing problem. We import from AI-creative-design community three methods: decomposition, co-evolution and emergence; we apply these techniques to combine elementary components in order to generate a fuzzy logic controller.