On the concept of fuzzy probabilistic controllers (Q797517)

This paper deals with a new approach to the construction of fuzzy probabilistic controllers where input-output data are treated as probabilistic sets. After summarizing some important features in section 2, probabilistic sets on finite spaces and their distribution representation with two decision rules (center of area method, maximal grade of the membership function method) are considered. In section 4, the construction of a fuzzy probabilistic control algorithm is embedded in the concept of probabilistic controllers where the probabilistic sets and membership functions are specified. The distribution function of the probabilistic set of control and the probabilistic relation make it possible to carry out moment analysis, in which the presented method distinguishes from respective constructions of controllers considered in previous papers. Some numerical examples illustrate the idea of the new approach, and specify the matrices of fuzzy controller (section 5). Finally, the authors emphasize that fuzzy controllers may be helpful in situations where experts have different control strategies but approximately the same goal of the control. However, the knowledge of the distribution function of the probabilistic set is required such that the need for reasonable methods leading to achievement of the distribution functions is stressed.