Reducing demand signal variability via a quantitative fuzzy grey regression approach

The total system performance of dynamic and complex supply chain networks depends mainly on accurate demand signal estimation as incorporated with an appropriate decision-making process. Due to the field of activity and architecture, however, it is hard to choose a proper forecasting and demand decision model that would befit the complexity of the system. This paper develops a conjoint intelligent hybrid system, comprised of an adaptive neuro-fuzzy inference system (ANFIS) based demand decision process, integrated with crisp grey GM (1,1) and fuzzy grey regression (FGR) forecasting models. We adopt this approach in an attempt to reduce the demand signal variability in supply-chain networks and to evaluate the system response to the proposed models under predefined, relatively low, medium and high demand signal variations. The results obtained from the simulation runs illustrate that the proposed hybrid system reduces the variability considerably; and also, could be considered as a substantial tool for reduction of supply chain phenomenon so called Bullwhip effect.