An identification algorithm in fuzzy relational systems (Q760403)

The author focusses his attention on identification problems in systems described by the use of fuzzy relational equations. A new algorithm is proposed for constructing models of fuzzy systems where the concepts of fuzzy discretization and clustering techniques are applied. The approach used by the author has two main advantages, viz. (a) reduction of the memory load in a computer implementation, (b) a unified treatment of fuzzy and nonfuzzy forms of information. Numerical examples make it possible to present the most characteristic features of the method and discusses its various applicational aspects. A possible way of generating a linguistic decision-making algorithm is discussed. Though the type of control rule is not precise it has quite high descriptive properties.