Author(s): G. Giuliano; A. Cancelliere; G. Rossi
Linked Author(s):
Keywords: Fuzzy inference systems; Multiobjective optimisation; Operating rules
Abstract: A methodology for deriving operating rules of a multipurpose reservoirs system through application of optimisation and fuzzy logic techniques is implemented. Two competitive objectives, namely the sum of the squared irrigation deficits and the municipal volumetric reliability, are taken into account. The ε-constraints method is applied to find a nondominated solution that represents the best trade-off between the objectives. Then a fuzzy inference system (FIS) that expresses the irrigation releases as a function of the reservoirs storage volumes is developed based on if-then rules that use the reservoirs storage volumes as premises, and the irrigation releases as consequences. Once the membership functions in the premises are defined, an adaptive technique is applied to calibrate the parameters of the conclusions of the fuzzy rules in order to model the data provided by the optimisation model. This leads to operating rules that can be finely tuned by decision makers on the basis of their own experience, by changing the membership functions in the premises. Results show that the performance of the adaptive fuzzy inference system well matches the output of the adopted multiobjective optimisation model, and therefore can be effectively used for real time operation.
Year: 2003