Quality classification of uranium dioxide pellets for PWR reactor using ANFIS

An ANFIS networks was used for classifying quality of green pellets for PWR reactor. It applied several physical indicators for classification; height; volume; weight and density. A total of 200 data sets of observation data was collected from one lot final compacting process of uranium dioxide pellets for PWR type reactors in the laboratory of experimental fuel element installation (IEBE) BATAN and used for training and testing the model. The performance criterion selected for the comparison between the actual and the estimated data are the root mean square error (RMSE), maximum relative error (MRE), and goodness of fit (R²). Up to 90% of the data could be correctly classified using this model. It is applicable in evaluation and classification of pellets quality.