نویسندگان
1 دانشجوی دکترای زمین شناسی گرایش آبشناسی، بخش علوم زمین، دانشکده علوم، دانشگاه شیراز
2 استاد بخش علوم زمین، دانشکده علوم، دانشگاه شیراز
چکیده
کلیدواژهها
عنوان مقاله [English]
نویسندگان [English]
In this paper, an Artificial Neural Network (ANN) is designed for the determination of unconfined aquifer parameters: transmissibility, storage coefficient, specific yield, and delay index. The network is trained for the well function of unconfined aquifers by the back propagation technique and adopting the Levenberg–Marquardt (LM) optimization algorithm. By applying the principal component analysis (PCA) on the training data sets the topology of the network is reduced and fixed to [3×6×3] regardless of number of records in the pumping test data. The network generates the optimal match point coordinates for any individual real pumping test data set. The match point coordinates are then incorporated with Boulton analytical solution (1963) and the aquifer parameter values are determined. The generalization ability and performance of the developed network is evaluated with 100/000 sets of synthetic data and its accuracy is compared with that of the type curve matching technique by two sets of real field data. The proposed network is recommended as a simpler and more reliable alternative for the determination of unconfined aquifer parameters compare to the conventional type-curve matching techniques.
کلیدواژهها [English]