Evaluating Seepage of Dam Body Using RBF and GFF Models of Artificial Neural Network

Dams have been always considered as the important infrastructures and their critical values are counted. Hence, evaluation and avoidance of dams’ destruction have a specific importance. Seepage occurrence in dams is an inevitable phenomenon. Despite all the progress in geotechnical engineering, up to now, seepage problem is the main conflict which occurs in dams. This study tried to estimate seepage of the embankment of "Boukan Shahid Kazemi’s dam” using RBF and GFF models of artificial neural network. To achieve this goal, the piezometric data set including 864 data were used. 70 percent of current data was used for training the network and 10 percent for calibration of two models. So 20% remained data was used for testing the network. Using suitable and applicable statistical parameters indicated that the RBF model with Levenberg Marquardt training and 4 hidden layers has high potential in estimating seepage, also the correlation coefficient for this model is 0.81 and the root mean square error was obtained 33.12%.