background and aim: the performance of water resources systems is very sensitive to uncertainty in model inputs. in this paper, adaptive neuro-fuzzy inference system (anfis), which is an improvement of fuzzy theories, is used to solve the optimization problem of reservoir operation. the main idea in this article is to consider hydrological uncertainties in anfis models based on grid partitioning, subtractive clustering and fuzzy c-means in dam reservoir operation and compare them using classical and fuzzy performance criteria.
method: it is a case study about the problem of simple operation of dez dam reservoir in single-reservoir state.
first, the problem of optimizing the use of dez dam reservoir with the aim of minimizing the lack of needs has been solved by using modeling training data. to solve the optimization problem, the ant colony optimization (aco) algorithm with a new mechanism called stochastic adaptive refinement (sar) has been used.
then, its results have been used to infer the behavior of the system in anfis and compare three methods of division. in order to compare the methods, in addition to the error criteria, the reservoir performance criteria have been fuzzified and compared with the classical performance criterias. finally, mamdani, sogeno and anfis fuzzy inference methods have been compared by using fcm.
results: fuzzy performance criteria indicate that fcm and subtractive clustering have improved fuzzy performance criteria compared to the optimal method in test data, and fcm has improved the objective function of the training and test data by 14% and 16%, respectively. the grid partitioning method has not achieved good results. sugeno's fuzzy inference has a better performance than mamdani's in inferencing system behavior, and anfis has also improved the objective function compared to sugeno's
conclusion: fuzzification of performance criteria provides more realistic criteria to professionals. fcm and then subtractive clustering are efficient methods in partitioning and generating fuzzy rules in inferencing system behavior. therefore, anfis based on fcm can be used as an efficient method for dam reservoir operation.