Development of a Self-Learning Approach

Systems based on fuzzy rules are widely used in the development of control, pattern recognition and machine intelligence systems. The structure of the fuzzy rule base is the most influential factor for the performance of fuzzy rule-based systems. This structure is defined by the number of fuzzy partitions and the formation relations used. The number and structure of the fuzzy partitions have to be defined during the design process. Thus, a large number of rules have to be generated. To avoid large computational costs, a reduction process is required. Current procedures for the design of fuzzy-based systems often do not take into account the signal or data specifications for the system considered. In the present study, a new approach for building a fuzzy rule-based system is developed. In this approach, design of the fuzzy rule base is based on the statistical properties of the data considered. A new framework is developed for automated and improved generation of fuzzy-based rules for identification and classification processes. Different benchmark data, comparative approaches, practical application , and hypothesis techniques are used to evaluate the effectiveness of this approach.