Seepage evaluation of an earth dam using Group Method of Data Handling (GMDH) type neural network: A case study

摘要

The seepage critical impact and significant destructive nature during and after construction of earth dams have been increasingly important topics during the past decades. In this paper, a new approach is presented for determination of seepage induced flow under and through an earth dam based on Group Method of Data Handling (GMDH) algorithm. After careful (detailed) studying of an earth dam called Fileh Khase dam located in Zanjan province of Iran, the permeability of soils was estimated by back analysis method using a Finite Element Method (FEM) software called SEEPW. Then, a number of 96 data sets were provided using SEEPW to use as a database according to allowable range of effective parameters such as permeability of clay core foundation of dam and water head in reservoir without any changes in geometry properties of the dam. This study addresses the question of whether GMDH type of artificial neural networks (ANN) optimized with genetic algorithms (GAs) could be used to estimate flow discharge through and under Fealeh Khase Dam. Results showed that GMDH type of ANN, provides an effective means of efficiently recognizing the patterns in data and accurately predicts the flow discharge through the Fileh Khase dam.