Investigation of the viability of modeling and estimation of Fe Fractions using Multi-layer Perceptron Neural Network

چکیده مقاله

Neural network models is a powerful computational technique for modeling nonlinear . This softwares used today for various forecasts in the sediments.The main objective of this study is prediction of spatial forms of iron through neural network software using some properties of sediment and the total concentration of iron.In this study, 180 samples were taken from superficial sediments of Reservoir Number 1 of Nimeh Well. After experimental measurements, a software technique of neural networks was used for prediction of the five bonds between the sediments. The total concentration of iron as well as the sediment properties were considered as the input, while each of the five bonds were regarded as output. To train the data, Levenberg-Marquardt algorithm was used, followed by consideration of a general MLP model for the prediction.Based on the obtained results and low values of the root mean square error and high correlation level, this model was able to perform a very good prediction, possessing a high congruence with experimental results

دریافت لینک دانلود مقاله