Accounting Measurement Uncertainty in Fuzzy Inference

Fuzzy inference systems are presently employed in the presence of uncertain models, or when deterministic models are too complex to be implemented. Input data to fuzzy inference systems are generally represented by crisp variables also when they represent experimental measurement results, thus disregarding measurement uncertainty. Since recent works have shown that measurement results can be effectively represented, together with their associated uncertainty, by fuzzy variables, this paper proposes a modified fuzzy inference system characterized by considering fuzzy variables as input data.