Institutional Repository
Technical University of Crete
EN  |  EL

Search

Browse

My Space

Predicting the spatial distribution of the aquifer head using a radial basis function network

Tsaparas Vasileios

Full record


URI: http://purl.tuc.gr/dl/dias/75312124-7826-4F39-9B29-784AEF4254ED
Year 2018
Type of Item Diploma Work
License
Details
Appears in Collections

Summary

The purpose of this work is to estimate the hydraulic head of an aquifer in the regional unit of Drama using an artificial neural network called RBFN (Radial Basis Function Network).Conventional modeling techniques are often time-consuming and costly and have limitations in data and knowledge. Artificial neural networks provide an alternative to these obstacles, as they can provide solutions without defining the relationship between the data and the results, by being trained by data and generalizing. Their function is based on the biology of the human brain and is a form of artificial intelligence.Artificial neural networks are trained by modifying the interneuronconnection strengths, known as synaptic weights, between their artificial neurons. This is done by feeding the network with input-output examples, to give them a desired response to input data and, thus, to modify their synaptic weights according to a learning rule. After a large number of iterations, the artificial neural network has constructed an input-output mapping and has adapted to the problem.In the study area we have data from 250 points from observation wells. For each point we have coordinates x, y and hydraulic head measurements. The neural network was trained on the majority of data and its accuracy was tested on a small number of testing data. Tests were performed by dividing the data set into subsets and training the neural network with each of them, as well as testswith random data. The accuracy of the neural network was examined in each case and graphical representations of the results and deviations from the measurement values were made.

Available Files

Services

Statistics