Abstract for: Fuzzy Modeling of Linguistic Variables in a System Dynamics Context
This paper builds on a previously proposed approach where fuzzy logic is used to incorporate linguistic variables in system dynamics modeling. The motivation for this approach is to include vague yet dynamic variables that are combined in a meaningful way. The essence of our approach requires the definition of membership functions as representations of the degree to which specific variable attributes hold, the application of a max-min direct inference approach as a way to combine two or more fuzzy variables, and the use of a defuzzification method that captures (summarizes) the joint effect of the linguistic variables. The objective of this paper is to study the implications of using two alternative defuzzification methods (largest of maximum and center of area) and to highlight various interpretation and modeling challenges associated with each defuzzification method. For illustrative purposes we use a variant of a sales and service model that is based on the concepts of product diffusion, backlog accumulation and personnel adjustments and their respective existing modeling representations in the literature. In summary, based on our findings, by substituting the Max-Min operator and eliminating inconsistencies among the fuzzy rules, the defuzzified values behave reasonably for both defuzzfication methods.