Compacting rules for fuzzy production system computation

Abstract

The authors analyze the inference process in fuzzy production systems which have linked rules. The process for obtaining inferences in fuzzy production systems described by a base of linked fuzzy conditional statements is characterized by a very high number of operations to be performed. During the development of the process, an important simplification may be achieved by means of a partial compaction of the rule base into punctual values. The values summarize the relationship between the linguistic values associated with the variables, which establish links between fuzzy conditional statements. As a consequence, an important reduction in the computation time and in the storage memory requirements can be achieved.<>